The US Government Just Shut Off Anthropic’s Best AI. (What Really Happened)

Summary

This episode of This Week in Startups opens with GRU Space (Galactic Resource Utilization) founder Skyler Chan in-studio, holding a brick made from simulated lunar soil. Chan lays out a plan to build the first hotel on the Moon using an on-site factory that mines lunar regolith with a robotic arm and binds the fine particulate with an Earth-shipped geopolymer — sidestepping the energy-hungry sintering and melting approaches. The pitch: the next SpaceX won’t build rockets, it’ll build habitats, and whoever figures out how to use resources off-Earth gets to own the land, Hudson’s Bay Company style. Chan frames a milestone path from making the first brick on the Moon to earning NASA’s ~$20B moon-base construction contracts, funded along the way by space tourism and $1M fully-refundable hotel deposits.

The back half turns to a live controversy: the US Commerce Department has ordered Anthropic to suspend access to its frontier Claude models — “Fable 5” and “Mythos 5” — for all foreign nationals over cyber-weapon and jailbreak fears, and Anthropic responded by shutting everyone out entirely. Jason Calacanis and Lon Harris debate whether this is a good actor self-regulating or political retribution against the one AI CEO, Dario Amodei, who never “bent the knee.” They dig into David Sacks’s account, an alleged jailbreak that Amazon’s Andy Jassy is said to have flagged, conflicting Politico and Axios reporting on a “universal jailbreak,” and the international backlash from a UK Labour MP and Canadian PM Mark Carney. Jason’s throughline: no one should have a single point of failure — you need a multi-model AI harness with dynamic routing.

The show closes with the results of TWiST’s $5,000 AI podcast-companion bounty. My AI Sidecast by Ricky Rivas wins $3,000 for fast, accurate live fact-checking, with Couchverse (Lemon Slice AI) and Convalenz (Beau Bratton) taking $1,000 honorable mentions — the latter surprising the hosts by turning a live read into presentation slides. Jason then teases the next bounty, Annotated.com, a permanent fair-use commentary layer for web content, and explains how he might incubate it inside his venture firm.

Highlights

”We basically got like manifest destiny going on on the moon”

“We should just be using resources on the moon. Like we built the US, did we bring everything from Europe? No, used the timber available in the US… NASA wants to build a moon base like ASAP, immediately. We basically got like manifest destiny going on on the moon.” — Skyler Chan, 12:30

yt-dlp --download-sections "*12:30-13:15" "https://www.youtube.com/watch?v=ZoK-prfL0sw" --force-keyframes-at-cuts --merge-output-format mp4 -o "manifest-destiny-moon.mp4"

Owning the Moon, Hudson’s Bay Company Style

“It looks like companies owning pieces of land on the moon, which, frankly, we’ve stated this in our white paper, is on our roadmap, is our aspiration is to eventually own a lot of the resources and the land on the moon. And we do this because we’re the ones who figured out how to even do stuff with it.” — Skyler Chan, 16:31

yt-dlp --download-sections "*16:31-17:15" "https://www.youtube.com/watch?v=ZoK-prfL0sw" --force-keyframes-at-cuts --merge-output-format mp4 -o "owning-the-moon.mp4"

Building the Hotel End-to-End Without Sending Humans

“We would build the entire hotel end-to-end without actually needing to send humans because we would have robots to assist us… For the first time in human history, you look up at the moon, like it’s no longer just a place to look at. It’s going to be a place like the Space Station where we’re going to have constant human presence.” — Skyler Chan, 18:00

yt-dlp --download-sections "*18:00-18:50" "https://www.youtube.com/watch?v=ZoK-prfL0sw" --force-keyframes-at-cuts --merge-output-format mp4 -o "robots-build-hotel.mp4"

”There is a beef between this company and the administration”

“There is a beef between this company and their politics and the administration and their politics. If you look at all of the parades of CEOs coming to Washington, Tim Cook, Michael Dell… There’s one that’s always been missing and that’s Dario and Anthropic.” — Jason Calacanis, 41:38

yt-dlp --download-sections "*41:38-42:20" "https://www.youtube.com/watch?v=ZoK-prfL0sw" --force-keyframes-at-cuts --merge-output-format mp4 -o "anthropic-beef.mp4"

”Regulate yourself before the government regulates you”

“If Grok, xAI, Google, Microsoft, Apple, OpenAI, any of them did the same exact rollout we would say thumbs up, thank you for being considerate. We don’t need to regulate you because you’re regulating yourself. And I have said over and over again — regulate yourself before the government regulates you.” — Jason Calacanis, 50:51

yt-dlp --download-sections "*50:51-51:35" "https://www.youtube.com/watch?v=ZoK-prfL0sw" --force-keyframes-at-cuts --merge-output-format mp4 -o "regulate-yourself.mp4"

”You cannot have a single point of failure”

“This is going to happen to Gemini, it’s going to happen to Grok, it’s going to happen to OpenAI, it’s going to happen over and over again. I do think this is going to drive people to realize you cannot have a single point of failure. That’s the key here.” — Jason Calacanis, 1:09:00

yt-dlp --download-sections "*1:09:00-1:09:45" "https://www.youtube.com/watch?v=ZoK-prfL0sw" --force-keyframes-at-cuts --merge-output-format mp4 -o "single-point-of-failure.mp4"

Key Points

• The next SpaceX builds habitats, not rockets (11:00) - Skyler Chan argues rockets are solved (per the SpaceX IPO); the next civilizational-scale company enables using resources off-Earth to build structures.
• How the lunar brick is made (12:30) - A payload factory excavates regolith with a robotic arm, filters for fine particulate, and binds it with an Earth-shipped geopolymer instead of energy-hungry sintering or melting.
• NASA’s ~$20B moon-base signal (15:00) - An announcement around GRU’s YC demo day that NASA would spend $20B on a moon base is framed as a huge demand signal.
• The TRL contracting path (15:10) - GRU wants to be NASA’s go-to construction contractor by proving Technology Readiness Level (1-9); demonstrating the tech on the Moon raises the odds of winning contracts.
• Business model: own the land (16:13) - With little hard regulation on the Moon, GRU’s roadmap is to own lunar land and resources, comparing it to the Hudson’s Bay Company holding Canadian land for 200 years.
• End state: Kardashev Type 3 (16:46) - The company’s ultimate vision is galactic-scale in-situ resource utilization (ISRU), elevating humanity on the Kardashev scale.
• Robots over humans (18:00) - Building end-to-end with robots avoids the cost of human life-support (ECLSS), power, and comms; the brick is the first thing humans lack “flight heritage” on.
• A third pillar: space tourism (21:35) - Beyond governments and billionaire-backed companies, Chan pitches tourism as a third funding pillar; the hotel “indexes” comms, power, and rover providers.
• The igloo: inflatable habitat + brick shield (24:52) - Bricks are stacked around an inflatable, expandable habitat to shield guests from radiation and micrometeorites.
• $1M refundable deposits (29:20) - Modeled on Virgin Galactic and Blue Origin, deposits are fully refundable and reserve a spot; buyers can also etch their name on the payload sent to the Moon.
• Moon as a stepping stone to Mars (35:04) - The umbilical cord to Earth must eventually be severed; using resources on the Moon and Mars is the only path to permanent off-world life.
• US government blocks Fable 5 & Mythos 5 (39:18) - Commerce orders Anthropic to suspend access for foreign nationals over jailbreak/cyber-weapon fears; Anthropic shuts everyone out to ensure compliance.
• The political read (41:38) - Jason argues there’s a genuine political beef: Anthropic’s talent leans heavily Democratic and Dario has never courted the administration, unlike other CEOs.
• Swap-the-names analogy (45:08) - Jason contends that if Google or Microsoft had done Anthropic’s cautious, monitored rollout, the reaction would be “thumbs up” — the actions are those of a good actor.
• Regulate yourself first (50:51) - Jason’s repeated thesis: labs should self-regulate before the government does it for them.
• The jailbreak dispute & Andy Jassy (51:54) - Per Sacks, a jailbreak turns Fable 5 into Mythos; Amazon CEO Andy Jassy (a ~10% Anthropic owner) is named as flagging it, fueling a “sandbagging” conspiracy theory the hosts reject.
• “Universal jailbreak” — he said, she said (1:01:19) - Politico reports Anthropic told the White House there is no universal jailbreak and the models are safe; the administration believes the tipster instead.
• Executive power and precedent (1:05:46) - Lon raises the First-Amendment-style concern of any president shutting off an AI model; Jason says the executive must hold emergency national-security power.
• The Iron Man analogy (1:08:06) - Jason likens the situation to the government telling Tony Stark he can’t have his suit — an unprecedented, fantastical scenario.
• International outcry and open source (1:09:12) - A UK Labour MP and Canadian PM Mark Carney both urge diversifying away from over-reliance on US models; China leans into open source.
• Build a multi-model harness (1:11:08) - Jason cites Nous Research’s Hermes and their own internal Kimi/Claude routing; own your harness (memory, skills, prompts, data) and dynamically route across models.
• Bounty winner: My AI Sidecast (1:28:06) - Ricky Rivas wins $3,000 for a fast, accurate live fact-checker; Couchverse (Lemon Slice AI) and Convalenz (Beau Bratton) get $1,000 each.
• Convalenz turns talk into slides (1:23:05) - Not really fact-checking — it builds accurate live presentation slides as you speak; Lon says he’d use it for his show.
• Next bounty: Annotated.com (1:28:53) - A permanent, fair-use commentary layer on web content to fight the “dead web” and silently altered quotes; Jason may incubate it inside his venture firm.

Mentions

Companies

• GRU Space (Galactic Resource Utilization) (11:00) - Skyler Chan’s YC-backed startup building lunar-regolith bricks and a robotic Moon hotel.
• SpaceX (7:44) - Its IPO frames the “next SpaceX” thesis and the 90%+ drop in launch cost.
• NASA (15:00) - The ~$20B moon-base commitment and the TRL contracting scale.
• Lunar Outpost (21:35) - Lunar Terrain Vehicle (LTV) provider Chan cites.
• Astrolab (Venturi Astrolab) (21:35) - Lunar mobility company named among LTV providers.
• Starcloud (18:00) - “Send a cube to space, then send thousands” analogy for scaling brick factories; also referenced re: space GPUs.
• Anthropic (39:18) - Maker of Claude Fable 5 and Mythos 5, at the center of the government shutdown story.
• Amazon / AWS (52:52) - Anthropic’s top partner and ~10% owner; alleged to have found and flagged the jailbreak.
• OpenAI (43:14) - Contrast case; Greg Brockman’s political donations and the exodus of talent to Anthropic.
• Google / Gemini, Microsoft, xAI / Grok, Apple (45:42) - Jason’s “swap the names” comparison set for a cautious model rollout.
• Nous Research (1:11:00) - Jeffrey Quesnelle’s lab; Hermes can use any backend model, the model for a portable harness.
• Mistral / Manus / Kimi (1:10:13) - Non-US and open-source model alternatives referenced.
• Palo Alto Networks (1:16:57) - CEO Nikesh reportedly said they found many security flaws in the model.
• Greylock Partners (1:13:56) - The VC firm through which Reid Hoffman holds an indirect Anthropic stake.
• Lemon Slice AI (1:18:28) - Maker of Couchverse, an animated-avatar podcast companion; honorable mention.
• Gamma (1:25:29) - Lon suggests it should acquire the live-slides idea.
• Plaud (7:44) - NotePin sponsor; Jason muses it could buy the live-infographics tool.
• Polymarket (2:53) - Prediction market Jason used to bet on the Knicks.
• Kith (6:03) - Designer behind Jason’s “lucky jacket” Avirex collab.
• Virgin Galactic / Blue Origin (29:20) - Space-tourism deposit models GRU studied.
• Hudson’s Bay Company (16:13) - Historical analogy for a company owning frontier land.

Products & Technologies

• Lunar regolith geopolymer bricks (12:30) - Fine regolith bound with an Earth-shipped geopolymer, the first product GRU aims to make on the Moon.
• TRL (Technology Readiness Level) (15:10) - NASA’s 1-9 readiness scale that gates contract eligibility.
• Kardashev scale / ISRU (16:46) - The company’s “galactic resource utilization” end-state framing.
• Claude Fable 5 / Mythos 5 / Opus (39:18) - The frontier models at issue; queries flagged as risky auto-downgrade from Fable to Opus.
• AI harness / dynamic routing (1:11:08) - Own the front end, memory, skills, prompts, and data, then route across any backend model.
• Hermes (1:11:08) - Nous Research model/harness that can use any backend LLM.
• Kimi 2.5 (1:11:09) - Open model TWiST tested in their internal routing setup.
• Plaud NotePin S (7:58) - Wearable recorder Jason uses to capture and summarize notes.
• My AI Sidecast (1:26:16) - Ricky Rivas’s minimalist live fact-checker; bounty winner.
• Couchverse (1:18:28) - Chrome-extension podcast companion with animated alien and cat-girl avatars.
• Convalenz (1:21:37) - Desktop app that builds live presentation slides from your speech.
• Annotated.com (1:28:53) - Fair-use clipping and commentary layer that creates permanent reaction pages for web content.

People

• Skyler Chan (11:00) - Founder of GRU Space, the episode’s guest.
• Jason Calacanis (0:41) - Host of This Week in Startups.
• Lon Harris (0:51) - Co-host and editorial director.
• Dario Amodei (41:56) - Anthropic CEO who never courted the administration; central to the political-beef read.
• Pete Hegseth (40:55) - Secretary of Defense whose tweet ties the policy to the Pentagon’s prior fight with Anthropic.
• David Sacks (41:08) - AI/crypto czar whose tweet is the source for the jailbreak narrative.
• Andy Jassy (52:52) - Amazon CEO named as flagging the jailbreak to the administration.
• Reid Hoffman (1:12:57) - Greylock/Anthropic investor whose stake and politics the hosts dissect.
• Greg Brockman (43:14) - OpenAI president cited for large political donations.
• Andrej Karpathy (43:14) - Cited as a top researcher who chose Anthropic.
• Bernie Sanders (1:28:00) - Proposed an “AI Sovereign Wealth Fund Act” 50% equity tax on major AI firms.
• Jeffrey Quesnelle (1:11:08) - Nous Research founder who came on TWiST with Hermes.
• Alistair Carns (1:09:12) - UK Labour MP who called for a homegrown AI industry.
• Mark Carney (1:09:12) - Canadian PM who flagged over-reliance on certain models.
• Ricky Rivas (1:26:16) - Builder of My AI Sidecast, the bounty winner.
• Beau Bratton (1:21:37) - Builder of Convalenz, honorable mention.

Surprising Quotes

“I love it when people are like, wait, so the moon hotel is just the first step? Like, that’s fucking crazy. It’s like the end state.” — Skyler Chan, 34:53

“So regulate thy self or thy product will get banned.” — Jason Calacanis, 51:27

“The only other time I can think of this happening was in Iron Man. A fictional story where they’re like, ‘Hey, wait, can you have that suit, Tony Stark?’” — Jason Calacanis, 1:08:06

“All of the open source models are using the frontier models to… like they’re distilling them and that’s how they’re training. So every new generation builds on the generation that came before anyway.” — Skyler Chan, 1:12:43

“Three billion parameters of the Condor.” — Lon Harris, 1:17:34

Transcript

Jason Calacanis: 0:00 He’s building the first lunar hotel in using an on-site factory that turns the moon’s soil into bricks.

Lon Harris: 0:09 You’re kidding me.

Jason Calacanis: 0:10 This is not a put on. You sure this isn’t a put on?

Skyler Chan: 0:12 We basically got like manifest destiny going on on the moon.

Jason Calacanis: 0:15 This is sci-fi becoming reality.

Skyler Chan: 0:18 We would build the entire hotel end to end without actually needing to send humans because we would have robots that assist us.

Lon Harris: 0:23 You’re asking people to put down a million dollar deposit to visit the moon, to own a moon condo. What, what do you get for a million bucks here?

Skyler Chan: 0:30 They’ve experienced everything that can be done on earth that they want to do something greater. The moon hotel is simply just the first wedge towards enabling us to build products on the moon.

Jason Calacanis: 0:41 All right everybody, welcome back. It’s Twist, it’s Monday, June 15th. The New York Knickerbockers have won the world championship. I was there, Lon. I was there when my brother Josh.

Lon Harris: 0:51 And in beautiful San Antonio, the Frost Bank Center.

Jason Calacanis: 0:55 Very simple ride from, you know, Austin, 70 minutes or so.

Lon Harris: 1:01 Yeah, not so bad.

Jason Calacanis: 1:02 Stop at the original Black’s BBQ.

Lon Harris: 1:05 Mmm, yes. For people who don’t know, the Black family of Texas, they’re having a war right now over who controls the barbecue. So there’s one young Black who runs the original Black’s BBQ that’s on the road from here to San Antonio, and then there’s Terry Black’s here in Austin, which is a different child of the Black family. It’s a whole thing.

Jason Calacanis: 1:28 We had a very similar thing in Brooklyn and Harlem. There’s a place, Patsy Grimaldi’s Pizza. Patsy’s Grimaldi’s brick oven pizza, fantastic. Frank Sinatra used to go to the one under the Brooklyn Bridge, and then there’s the one up in Harlem. Anyway, long story short, same situation, you know, the kids, the cousin, the uncle, whatever, everybody’s got the same last name.

Lon Harris: 1:46 And they all want to be the inheritor of the tradition because Black’s one of the oldest barbecue places in central Texas.

Jason Calacanis: 1:53 And it was fantastic. Really great beef rib, really great brisket. If I were to swap those two items with Terry Black’s, I probably wouldn’t know the difference if I’m being honest, but I just want to say, you know, my brother had an observation being from Brooklyn. He’s like, wow, Texas is amazing. Everybody’s so polite, everybody’s so cordial and it’s like it used to be in Brooklyn when we were growing up. You just assume everybody’s in good faith and we walk into Black’s BBQ, what is that, Laredo? I don’t know, Kyle, Buda, somewhere down there. Anyway, there’s like two or three down there. And everybody’s like, oh, hey, you going to the game? I said, yeah. They said, oh man, good luck man. I said, oh, thanks. You a Knicks fan? They said, no, I’m a Spurs fan. I said, oh, really appreciate that, you have a great team and just really cordial and I was six rows behind the Knicks bench. Cost me a fortune, but you know I PolyMarketed YOLO’d 6k into it in 1:15, so I paid for the ticket.

Lon Harris: 2:52 Gambley.

Jason Calacanis: 2:53 I love these prediction markets. It’s so much fun. Shout out to my guys at PolyMarket. I think I saw Shane with courtside seats and

Lon Harris: 2:59 Probably.

Jason Calacanis: 3:00 Everybody was there. It was fantastic and just a great relief after, you know, I was there for two of the failed attempts versus Houston and versus Spurs. I was there when the Spurs won on our court. Spurs got a bright future, a lot of great young players. But man, this Knicks team is transcendent. Thank you to Leon, thank you to Coach Brown, thank you to Jalen Brunson, OG Anunoby, Mitchell Robinson, Josh Hart, I mean just Bridges, F&M picks, just so many great players and had a great time crawling in this season to Knicks Fan TV. Shout out to my guys over there Alex, CK2K, and of course CP the Franchise. Great YouTube channel to follow. They do this Fan TV, which is like a thing in soccer, where they have all these fan shows. The fans just get on and, you know, they’re drunk and blah blah blah and they’re yelling, but they’re obsessed with the game. KFTV, Knicks Fan TV, has got 100,000 followers and he’s been following the team for nine years. What a great story. He was following the team when they had 17 wins a season. There’s 82 games a season, Lon. When we were the worst team in the league, he was following this team and had literally 30 people watching his live stream, you know, then when they win games he has 3, 4, 5,000 people watching live and then another, you know, 100,000 watching after that. So, just great all around. Amazing for the city of New York. Couple of notes. Couple of notes. Number one, during the national anthem, put your hand over your heart, take your goddamn hat off, and no screaming. I want you to wait to the end and then scream.

Lon Harris: 4:34 People were screaming?

Skyler Chan: 4:35 Well, I mean a little bit of… I think the one moment you’re allowed to get excited during the anthem is ‘and the rockets’ red glare’ when they hit that note, you’re allowed to give a little… you’re allowed to give a little woo, a little cheer, yeah.

Jason Calacanis: 4:48 Under… not for me, but okay. I’m okay with it. It’s kind of a tradition.

Lon Harris: 4:53 Under… That’s a tradition. Yeah, like ‘oh, they hit that, they hit that note, the… well done.’

Jason Calacanis: 4:58 Okay, fine. Yeah, it’s kind of like a roar in the end, it’s like a pre-applause. But there’s guys who are yelling ‘go Knicks’ like every time and the person’s trying to sing the national anthem. It’s not… it’s uncouth.

Lon Harris: 5:11 Disrespectful.

Jason Calacanis: 5:12 Second thing that’s uncouth. Uh, you know, if you lose, which I’ve spent 50 years doing with this team, you can’t take it out on the other fans. There was like a little random violence. I’m also not down with the vulgarity and the burning of other people’s jerseys. It’s not cool. Be classy in victory and be honorable in defeat. Right. This is just a PSA for young fans. You know, I don’t mind if you climb a lamp post or whatever, but I’m not down with you burning a, you know, a fire engine or a police car, none of that shenanigans. Keep it classy, folks.

Lon Harris: 5:40 Yeah, people get excited. It seemed like I saw a lot of footage of these celebrations in New York and it seemed like the rowdiness was an outlier. Most people, it felt like it was a real community, all in good fun sort of vibe.

Jason Calacanis: 5:53 Now I’ve been on the road, I think this was my ninth game in this postseason, and then one time… yeah, a couple times last year, but just for this postseason… season. This jacket. I…

Lon Harris: 6:02 Oh, the lucky jacket, yeah.

Jason Calacanis: 6:03 I just want to say thank you and shout out to Kith. I had to pay for this, it’s not a freebie or anything, but this is an Avirex, which is like the classic jacket. And they did this, um, this incredibly designer, K-I-T-H, Kith is an incredible designer. This jacket costs a pretty pen, I don’t want to say how much, but you know, a couple G’s. It’s now become like a classic, you can buy, like people are selling these in the aftermarket. I think they only made like a hundred of them. Guys like, ‘Hey, you know, we’re huge fans of yours.’ I was like, ‘Okay, great, can I get this jacket?’ And they’re like, ‘Oh yeah, well here’s how it works. You go to Williamsburg, you line up at 5:00 AM, and then you’ve got a pretty good shot, they give you a ticket.’ I said, ‘Hey, let me stop you right there. I’m old and I’m not going to Williamsburg. If it was a 25-year-old J-Cal, 30-year-old J-Cal, I’m all in. Your fans, I’m fans. How about this? You send me an invoice for five dimes. I send you five dimes. You sit on my five dimes, put it in your interest-bearing account, make 500 a year off me, whatever it is, you know. And then when you drop the Knicks/Heat, I’ll just tell you which ones I want, you maybe send it to me, whatever, make a little VIP program.’ They said, ‘That was very interesting,’ and sure enough, I get an invoice, I get the jacket, boom. Jacket, nine for nine in playoff games. I went, the Heat… the air conditioning doesn’t work in the dump of an arena the Spurs are still in. They’re getting a new arena.

Lon Harris: 7:21 The Frost Bank Arena is where the Spurs play.

Jason Calacanis: 7:24 I mean, just literally tear it down, it’s, I mean, this thing is, it’s rough. Anyway, they’re going to build a new arena, modern. But they don’t have air conditioning working in this building.

Lon Harris: 7:34 Wow.

Jason Calacanis: 7:35 And so it’s hot down there, and they just, it gets hot in there. I wouldn’t take the jacket off. I took it off a couple times just to hold it up, but…

Lon Harris: 7:42 Well, it’s lucky, the team’s counting on you, you know.

Jason Calacanis: 7:44 I’m sweating and, anyway, you know we got a lot on the docket today. I’m really excited about today’s guest because space is so big with the SpaceX IPO. I just don’t want to miss a moment, so I gotta just get this Plaud pin on here.

Lon Harris: 7:56 Yeah, get the Plaud pin on.

Jason Calacanis: 7:58 I press the button, I now have the haptic going, now everything’s recorded. You know me, I’m on a tear right now. I’m on a tear right now working at the company. You see behind the scenes all the stuff I’m doing, I’m vibe coding this new Claude, the Perplexity, I’m like I’m doing a lot with Perplexity and Claude to automate all this stuff and level everybody up. We got five associates in training starting, so I’m really excited about our business and running it and you know all the great stuff we’re doing. But I don’t want to miss a minute, so I use Plaud. I went on a hike the other day on the ranch, had a lot of action items that come to my mind when I’m hiking in nature with my 25-pound kettlebell, doing my old-man rucking. I never miss anything because I just turn the Plaud on. I might not talk for 15 minutes, and then I talk for 10. Then I’m quiet for 15, I talk for another 10. Get back, put it in the cradle, Cat’s in the Cradle, boom. Summarizes it, I get the bullet points, I pull it up on my desktop.

Lon Harris: 8:51 Yeah.

Jason Calacanis: 8:52 I’ve never touched my phone, by the way. And then I have all my notes waiting for me there, organized. And then I can share you the raw feed with all those notes or, you know, I just look at it, I remember a couple things, cut and paste.

Lon Harris: 9:00 thing right into Slack and I’m ready to go.

Jason Calacanis: 9:03 Yeah, the app.

Lon Harris: 9:04 (sponsor read) Don’t miss a moment. Pulls the whole thing together. You could put the app wherever you want, and then all your conversations just live across all your devices. So if you want to make your work more efficient like J-Cal, you need to get your own Plaud NotePin S. If you check it out at plaud.ai/twist and use the code TWIST, they’re going to give you 10% off your purchase.

Jason Calacanis: 9:23 Mm.

Lon Harris: 9:24 All right, let’s get started. We have a great guest today. I think we’re going to get right to it. Are you going to do news and then?

Jason Calacanis: 9:28 No, no, I think we should get right to our guest. He’s building the first lunar hotel using an onsite factory that turns the moon’s soil into bricks, just like this one I’m holding up in front of you. Please welcome Skyler Chan of Gruv Space.

Lon Harris: 9:45 Now he sent us bricks!

Jason Calacanis: 9:46 This is not a put on. You sure this isn’t a put on? Lunar soil? We set them.

Jason Calacanis: 9:51 (sponsor read) One of the most important things you can do as a busy founder is stay focused. You need to prioritize the decisions and workflows that are going to move your company forward, not get lost in extraneous details and minutiae. That’s why the best decision you can make for your business is signing up with Every. Every is the all-in-one platform for incorporation, banking, payroll, benefits, taxes, basically your entire back office. They’re going to keep up with all the bookkeeping and paperwork so you can stay laser-focused on your next big product or hiring, features, customers, all that important stuff. If you’re just getting started, they’re going to handle your Delaware C-Corp and getting yourself a registered agent, all with no legal fees and no delays. And Every has worked with over 1,000 startups already. That’s the kind of know-how and experience you can rely on. So, head over to every.io and stop wasting time on stuff that isn’t growing your company. That’s E-V-E-R-Y dot I-O.

Jason Calacanis: 10:47 It’s not a put on. This is like for real. It’s a real startup planning on building a lunar base.

Lon Harris: 10:53 Explain how you came up with the idea and where you’re at in this journey because this is sci-fi becoming reality.

Skyler Chan: 11:00 Thanks. Yeah, thanks for having me on the show. So yeah, I’m Skyler. I started this because I always wanted to become an astronaut when I was a kid. So obsessed with becoming an astronaut as a kid that I decided that why don’t we all go to space? And so if we have this end-state future of exciting like first cities being built on the moon and Mars, like the next San Francisco, the next New York on the moon and Mars, the question is working backwards from that, what are the series of civilizational-scale technologies that we need to develop to make this happen? It’s not just rockets anymore. Clearly with the IPO, we saw that ship has sailed. The next SpaceX is going to be building the first habitation modules, the enablement of using resources off-earth, but using resources on the moon to create structures so we don’t have to pay for all the shipping from Earth to the moon. And so that’s kind of where we’re at. We just went through YC last batch and moved into a big space in the center, shipped to the moon. So it’s pretty exciting.

Jason Calacanis: 11:59 So, maybe we can get into technically what you’re actually doing. We understand Moon base, great. We also understand, hey, shipping bricks and materials from Home Depot to the moon is quite wasteful. So you’re going to build some sort of a factory, got to ship something up there, but then I’m seeing here in this video, very evocative, you have designed and tested building these bricks out of moon soil? Is it soil? Is it what what is the moon constructed of when we would compare it to soil on Earth? And then take us through this process and how do you test something like that? Do you actually have some moon rocks that you bought for a billion dollars to test it? Tell me everything.

Skyler Chan: 12:30 Yeah, let’s dive into details. So basically what we’re doing with this lunar regolith, and lunar regolith is topsoil on the lunar surface. So think of it just like dirt all over the moon. The idea here is that like you said, it doesn’t make sense for us to just keep shipping shit to the moon because it’s very expensive. And we should just be using resources on the moon. Like we built the US, did we bring everything from Europe? No, used the timber available in the US. And so what we’re doing is we’re building a factory, think of it like a payload, like a mini spacecraft that we’re sending to the lunar surface. And what it’s going to do is it’s going to have an excavation stage, a couple stages, excavation stage on the moon where we mine lunar regolith with a robotic arm, then we filter out the lunar regolith. You want the fine particulate because it has more surface area to bind it with a geopolymer which we bring from Earth. Now if you just ChatGPT like how do you make stuff on the moon, right, you’re going to find sintering and melting among other proposed approaches to do this. The reason why we didn’t choose these approaches is because of energy. And when we think about the sort of timeline that we’re doing with, right, NASA wants to build a moon base like ASAP, immediately. We basically got like manifest destiny going on on the moon. And so in terms of a timeline, like we can do things like laser the ground with like thousands of degrees of Celsius, it’s too costly and we don’t have the equipment. So the decision we made at Argo is like let’s use a geopolymer-based approach with a robotic arm, with some other stuff in our payload that we’re going to reveal soon to use the resources on the moon intelligently so that we can bring down the cost of construction on the surface.

Jason Calacanis: 13:26 Got it. Now your hope, I’m guessing, is that NASA gives you a contract to be the contractor to make the bricks that make the lunar base, or do you want to be the owner of the lunar base? Are there going to be multiple lunar bases? Because it sounds like building a lunar base is a multi-billion dollar product, and it sounds like you solved one of the acute problems, but I don’t know if they’re going to just let some random YC founder have a base on the moon. I know this is like not regulated, but there’s going to be some regulation going on at some point of this. So talk a little bit about what your aspiration here is and who your customer is. Obviously your customer right now is venture capitalists who are crazy and want to, you know, do this Hail Mary amazing idea, awesome. But you got to get a customer at some point, and the customer’s got to pay for something. So talk about how what the business model here is.

Skyler Chan: 14:55 Yeah, so the exciting thing is like pre literally on our demo day for YC was when I mentioned an event where they basically announced that they would spend 20 billion dollars to build a moon base, which is a great demand signal for us. So now we’re talking, like, the timeline post that time frame. Um, yeah, the short answer is, like, we want to be NASA’s go-to moon base construction contractor. And the pathway to getting there is to prove something called TRL, right? So, Technology Readiness Level. NASA ranks you on the scale of 1 to 9. The higher you are, the more likely you are to get contracts. And the best way to do that is to demonstrate that the tech actually works on the moon, which is why we’re doing this mission to make the first-ever product, the brick. But the second half of your question is, like, the business plan, which is very exciting. To preface this all, it’s a very blue ocean market. As we’ve seen with SpaceX, in the early days, it wasn’t obvious that, like, reusable rockets was the play. It wasn’t obvious that Starlink was going to evolve. Just like with the moon, there’s a lot of things that are not obvious yet for why construction, more specifically, the ability to use resources off-Earth, decoupling us from Earth, is going to lead to, like, this massive explosion in terms of value. And so for us, it looks like, of course, the moon holds out simply just the first wedge towards enabling us to build products on the moon, do the demo on the moon, then earn the moon base construction contracts and then scale, right? And to your point as well, regulations… it’s very interesting. There’s actually not much set, hard, clear-cut regulation for who does what on the moon. So our prediction is it’s going to look like, if you look at the Hudson’s Bay Company, right, in Canada, they owned a huge chunk of land for 200 years before they sold it back to Canada. And so it looks like companies owning pieces of land on the moon, which, frankly, we’ve stated this in our white paper, is on our roadmap, is our aspiration is to eventually own a lot of the resources and the land on the moon. And we do this because we’re the ones who figured out how to even do stuff with it, right? Because we’re the resource utilization company. So that’s the end-state vision. After that, eventually, it’s really to hit the final form of our company, which is galactic resource utilization, galactic-scale ISRU, elevating humanity to Type 3 on the Kardashev scale. So that’s really the end state of our company.

Jason Calacanis: 17:04 Yeah, wow. That’s… it’s interesting. I’m curious, how far down the road do you need… can you get without humans going to the moon? Like obviously, setting the factory, getting started, that’s all going to be sort of automated.

Skyler Chan: 17:18 You mean a few astronauts to help construct?

Jason Calacanis: 17:19 You know, that, like, even if it’s the final stage of the process before guests start arriving or what… like, at some point, you’re going to have to have humans going to the moon.

Lon Harris: 17:37 This is pre-concierge. Pre-concierge at your lunar hotel.

Skyler Chan: 17:42 Yeah. Yeah, well, that’s the thing is, like, everything that we can do with robots is a lot cheaper than having to send humans, right? Having to send humans up to the moon, there’s a lot of other associated costs, like life support systems, ECLSS, etc., etc. Power, comms, etc. So the idea is… We would build the entire hotel end-to-end without actually needing to send humans because we would have robots to assist us. This is important because if you think about the tech tree in terms of like how you sequence this, we’ve done robotics on the moon and Mars. We have something humans have something called flight heritage. What humans don’t have flight heritage on is how do you use resources on the moon to make a product, which is why we decided in our roadmap for our company is like okay, let’s laser focus ourselves on proving that we can make a brick on the moon, right? That is going to unlock all these other things, because what’s the next stage? Similarly to other companies that, for example, like Starcloud, they send a cube to space, what’s next? Let’s send like thousands of these bad boys up. For us, okay, we can prove that we can make a brick on the moon, let’s send thousands of these bad boy factories up to the moon to just crank out more bricks, just accelerate the construction. This is a very exciting future at the end of the day, because for the first time in human history, you look up at the moon, like it’s no longer just a place to look at and people visit. It’s going to be a place like the Space Station where we’re going to have constant human presence on the moon, which is a very exciting time to be alive.

Lon Harris: 19:12 Yeah, and all this is enabled Lon because Elon has taken the cost of getting to space, I think down 90% plus now?

Skyler Chan: 19:21 Yeah, if you see those charts, it’s exponential. As soon as you can reuse the rockets, the costs just plummet.

Skyler Chan: 19:27 (sponsor read) Debugging sucks and it takes up time your team could be spending on awesome new features and products. But now there’s a better way. Sentry’s AI-powered debugging agent, Seer, isn’t just guessing about what might have gone wrong with your system. It’s analyzing your actual data and because it has context, it spots buggy code before that code ruins your entire day. Plus, Sentry works alongside coding agents like Cursor, passing along an idea for a fix that can then be applied to your codebase and sent to a human team member for review. End-to-end automation, all the way from bug detection to pull request. So join the millions of devs and companies like Claud and Disney Plus who use Sentry to move faster. Check them out at S-E-N-T-R-Y dot I-O slash twist and use the code twist for $240 in Sentry credit.

Jason Calacanis: 20:11 We’ll get that famous chart and pull it up in a second. And so but then we have like lunar rovers and those kind of things which have robotic arms on them, but you are going to need some providers for your company, Skyler.

Lon Harris: 20:26 You’re not going to be able to make Optimus robots or you’re not going to be able to make the robotic arms, so is there a company making robots that work in space or is that all been limited to government now? And who and who are they? Because that’s the exciting part about this for me, Skyler, is so many entrepreneurs now because you can get to space for ten times less are now going. So really interesting if we can get there for ten times less, now we’re going to be need a thousand different problems to be solved. thousand different problems to be solved, 100,000 problems to be solved. And one of them is, you know, will this GPU work in space? So NVIDIA’s working on a specific GPU for space because it has—and we had Star Cloud on This Week in AI the other week—they have to make it so their GPU, like if you just send up an H100, it might come apart. It’s not made for the G-force going to space. It might not have the heat profiles and be designed to dissipate heat the way it does in our atmosphere. So man, there’s so many different things that have to occur.

Jason Calacanis: 21:29 Who’s the company making the robots that actually work in zero G’s in space for radiation, etc.?

Skyler Chan: 21:35 Yeah, there’s a couple companies out there. I mean if you look at the LTV providers, so LTV is like the Lunar Terrain Vehicle, you’ve got like Lunar Outpost, Astrolab, there’s a whole other set of companies. You know, we’re obviously in the mix of all these different folks and we have different things coming up, which we’ll save that for later. But yeah, there’s a whole host of group of companies there. There’s a lot of like robotic arm companies as well that do construction or welding and have done demos on Earth. Um, there’s a lot of comms companies, there’s a lot of power companies. So some bit of background, this is important I guess, is when you look at the lunar economy, you’ve got a lot of different surfaces, right? Surfaces for like providing power, providing comms. Like power as in like nuclear, crazy stuff on nuclear to like solar to like laser beaming on the moon, like some really cool shit. And you got this whole like bucket of people here. You got the comms guys, so how do you like FaceTime your friends on the moon or back home? The nice thing here, and rather I think the bigger takeaway is this: if you think about the space industry as a whole, it’s really got two pillars underpinning it, right? You’ve got governments and billionaire-backed companies. What we’re doing at Acrux is we’re taking the unconventional approach and we’re saying, ‘Hey, let’s use a third pillar: space tourism,’ extend it to literal moon and build the first hotel on the moon. The hotel is actually nice because if we win, it’s like an index on everyone else because comms guys, well we’ll be using them. The LTV rovers, well they won’t have to be just depending on NASA to stay alive but like we can use the rovers for ATBing on the moon for tourists etc., etc. Um, so it’s really this whole ecosystem, right? The moon hotel, interestingly enough, again, key distinction is the branding is distinct, hotel and base, but the key technology is really the same. It’s the enablement of using resources on the moon for building all kinds of products.

Jason Calacanis: 23:22 So, so tell me about this hotel in space. Is that like actually part of the vision right now and, you know, obviously you have customers. NASA might just say, ‘Hey, yeah, we’ll give you a contract to make us 10,000 bricks,’ and here’s the funding and here’s your milestones. Oh, that’s all great. But talk to me about this like really far-out-there concept that, hey, if you go to space, you might need a bed. You might need a hotel room.

Skyler Chan: 23:45 Yeah, yeah. I mean, it’s inevitable. I think one nice thing, call it what we got was, you saw on the Elon JPMorgan summit, like they were talking about moon hotels too and so, yeah, it’s on the roadmap. Like if you think about critical path, so moon hotel, like technically like the— On its core, what does that look like? You’ve got, let’s say, a habitat on the moon.

Jason Calacanis: 24:03 Okay, let’s break that down further. Let’s work backwards. Let’s say all of us, you know, Jason, Lon, etc., we just teleport to the moon right now. We take off our spacesuits. What’s the first thing that’s going to kill us?

Skyler Chan: 24:16 The lack of pressure on the moon.

Jason Calacanis: 24:17 Yeah.

Lon Harris: 24:17 We’re just going to float off into space.

Jason Calacanis: 24:19 Just exploding.

Skyler Chan: 24:20 Yeah, it won’t be pretty, right? Temp swings, radiation, and then the potential for micrometeor impacts. So then what we did at Gru was like okay, there’s a lot of other problems obviously like okay how do you recycle the water, how do you recycle this and that but these things have been solved again on the Space Station. Let’s focus on what’s critical about like at its core. So if you’ve got pressure and temperature, like working back from that now, what do you need to create as a product to solve for that? Well, we need to have an inflatable like expandable habitat. Okay, the last two problems of radiation, micrometeorites like again how do you think about that problem? Well, it turns out if you can use the resources on the moon and make not just bricks, but all kinds of material using the resources on the moon. Bricks are just the first step to shield the habitat so that you have an igloo-like structure where you have the inflatable in the middle and you’ve got the bricks placed around it like an igloo. That would protect you from the radiation from outer space and also the potential for micrometeorites on the moon. And so what we were like is okay, if we want to open up this hotel, like we have to actually prove that this stuff works on the moon and also on Earth along the way, but more importantly, it’s very visceral we have proof that this functions on the moon, it’s like it’s legit. And so that’s actually why how we ended up doing this milestone path that’s laid out on our white paper of let’s build a brick on the moon and let’s prove that the inflatable can expand.

Jason Calacanis: 25:45 And I’m showing on the screen here for those of you who are not watching and you’re listening, you can go to GRU space or gru.space is the domain. And you’ve got this really nice timeline here, you know, it starts with hey, you get your first lunar systems there and you test making the bricks, then you have this cave base, very basic dome, then you have the first lunar hotel and you’re seeing the bricks go around it and then you have picked an architecture style that is I would say neo-Greek Lucas Star Wars aesthetic.

Lon Harris: 26:26 Yeah, it’s got a Death Star vibe.

Jason Calacanis: 26:28 Death Star vibe.

Skyler Chan: 26:29 No, I think this is New Republic style actually. Death Star would be a little more military.

Jason Calacanis: 26:35 The cave one, the cave one looks like a Death Star.

Skyler Chan: 26:37 Yes.

Lon Harris: 26:38 The cave one definitely has Death Star vibes.

Jason Calacanis: 26:40 This one you’re right is more like Naboo. This is Naboo. This is Princess Amidala. This is Renaissance, which by the way, Lucas there’s a building you pass by in Marin that’s like a civic center or something and you see it when you’re when you’re going north it’s on the right, when you’re going south to the Golden Gate Bridge it’s to your left, you can barely see… But that’s where he got that Nobu architecture style from, Lon pull it up in a second here, one of the producers will, but talk to me about picking the style of what this would look like because that does not seem the most utilitarian, the most basic, efficient, you’re doing some flourishes there. You’re making it look elegant and interesting and sci-fi and dare I say inspiring. By the way, here’s that building. What is that building called Lon?

Lon Harris: 27:27 Uh I believe this is the Marin County Civic Center.

Jason Calacanis: 27:30 Yes.

Skyler Chan: 27:31 Yeah.

Lon Harris: 27:32 In San Rafael, San Rafael. Pretty amazing.

Jason Calacanis: 27:37 Is this the one that was Frank Lloyd Wright or am I crazy?

Lon Harris: 27:41 I mean, I believe it was. Yes, I am correct. This was designed by the great architect Frank Lloyd Wright.

Jason Calacanis: 27:46 Alright, let me pat Lon on the back here. Very good. Alright, so talk to me about the aesthetics here because you’re asking people to put down a million dollar deposit to visit the moon, to own a moon condo. What do you get for a million bucks here?

Skyler Chan: 27:59 Yeah, so okay, two things. First of all the style, right? And so you’ve basically got it bang on. It’s when we thought about the style, like we got to think about the zoom out from our own like little sliver of 80 years or maybe longer of life in this universe, right? When thousands of years later we look back at the first structure, man-made structure on the moon, like it has to be pretty—it has to be pretty sick, you know? Like it has to be—it would be a good benchmark. It is like the first bastion of like humans’ mark on the moon. I mean, that is—that’s really exciting. And so the thought process was like, you know, obviously looking at the wonders of the world that already exist, but also like what would something that looking back, right, our great-great-great-great grandkids would be like, whoa, like yeah, like my like ancestors worked on this shit. Look at that.

Jason Calacanis: 28:43 Like the Acropolis, the Parthenon.

Skyler Chan: 28:45 Yeah, exactly. Right, yeah. So it was a bit of Greco-Futurism and and uh actually it was a trip I did when I was at still at Berkeley as a student there. I went to Palace of Fine Arts with my parents and it was like I was just like whoa like imagine this on the moon. This was before Drew start but that was what—ironically my mom’s an architect and she thought the design was shit. I still think it’s good but we got featured in architecture magazines and I was like see mom.

Jason Calacanis: 29:04 Love it. I love it. Now what uh psychedelic were you on? You on Psilocybin or DMT when you came up with this idea when you were at Berkeley? I kid, I kid.

Skyler Chan: 29:16 Haha.

Jason Calacanis: 29:17 But talk to us about the million dollar deposit.

Skyler Chan: 29:20 And so yeah, about that. It’s so basically the thought process behind this as well was with okay, in order to ensure that people have a spot here, right, how should we, you know, work this out? And we studied Virgin Galactic, we studied Blue Origin, but we also studied ones that didn’t work so well. One important thing is is that we designed a—they have to be fully refundable, right? It’s only fair that way. But more importantly, this is a way so that people can reserve their spot in advance prior to the hotel being opening. Uh frankly, a big piece of this of like why do people do space tourism, right? Let’s think about that. Why do people even want to do something as crazy as this? And it’s not for everyone, like let’s be clear it’s not for everyone. One of the biggest things is because they’ve experienced everything that— can be done on Earth that they want to do something greater. They want to feel like they’re a part of something even bigger than themselves. It’s about legacy building. And so it’s an opportunity for people to be a part of that. Um, one of the nice things that we’ve actually, um, are going to start to offer is that our payload that’s going to the lunar surface, we’re obviously making a brick on the moon, but there are opportunities for people to etch their name or their company on the payload that goes to the moon and they’ll be there forever.

Jason Calacanis: 30:22 Cool.

Skyler Chan: 30:23 Yeah.

Lon Harris: 30:27 Oh, a little marketing.

Skyler Chan: 30:29 A little marker there.

Lon Harris: 30:30 (sponsor read) As a founder, you’ve taken the time to become an expert in your niche, and you fully understand the entire landscape. But with so many regulations and rules that you need to follow from federal agencies and offices here in the US to the European Union and beyond, it’s impossible for even the most diligent person to keep up with every exacting requirement. That’s why you need to simplify compliance with a trusted partner like Vanta. All my startups use Vanta, and they love Vanta’s AI-powered platform that automates your entire compliance process. Whether you’re preparing for a SOC 2 or you’re running an enterprise GRC program or you’re doing an audit, Vanta’s going to worry about the security so your team can focus on building great products. That’s why some of our favorite companies like Ramp and Writer are spending 82% less time on their audits by working with Vanta. Whether you’re a fast-growing startup or a global enterprise, Vanta is here to help you automate. Vanta is here to help you automate your security and compliance and earn and prove trust. So get started today at vanta.com/twist. That’s vanta.com/twist.

Skyler Chan: 31:36 I don’t know if you can see, Jason, your name is etched into this brick.

Jason Calacanis: 31:40 Oh, is it really?

Skyler Chan: 31:41 It’s a little too small to see. I’m going to have them zoom in here for a second so you can see it, but they’ve etched Jason Calacanis into this brick.

Jason Calacanis: 31:47 I love this. Leo Laporte when he did the TWiT brick house, his concept was you for I don’t know if it was a $100 or $500 you could support independent media like This Week in Tech, TWiT, and you could buy a brick in the brick house. And then when you came to visit the brick house, you could, uh, find your brick. It was a really genius idea.

Lon Harris: 31:58 Oh look at that, G-R-U.

Skyler Chan: 31:59 Yeah, there it is, yeah.

Jason Calacanis: 32:01 Amazing. All right, and so you’re going to keep these deposits. You get the interest on them, put them in a nice five, six percent interest-bearing account, you sweep the interest. That’s great. Fair enough. And then I get, what am I getting for my million dollars? Is it like a, I get to stay in a hotel for one night, ten nights? Do I own a piece of it?

Lon Harris: 32:27 What do you want to be on the moon for ten days? What are you going to do up there?

Jason Calacanis: 32:32 I mean, if I took the time to go and took the risk, yeah, I want to get a full no. I mean, I don’t think…

Lon Harris: 32:38 That’s a good question, like for Jason, would you want to go to the moon?

Jason Calacanis: 32:41 I’m going. I’m going to go. Um, well, a friend of mine’s got a rocket ship company and we talked about it. So I think when he goes up, uh, you know, if he wants to have a couple of his mates with him, I might get a, I might get a shot at that. Um, so when he thinks it’s okay to go up, that’s maybe when I would decide it’s okay to go up. Uh, but I don’t want to be, you know, I got three daughters. So I think… The power move is to wait until… yeah… you know, they’ve done… Yeah, I think maybe if there’s been like a hundred trips to the moon or something, I’m going to go.

Lon Harris: 33:10 Yeah, you don’t want to do the Titan submersible first trip. You want to wait for like a hundred trips and it’s like, okay.

Jason Calacanis: 33:15 What’s interesting you bring that up, Lon. It’s… there are people who want adventures who have enough money to kill themselves. We call it toxic wealth in Silicon Valley where people, you know, go buy some hyper sports car and because they made a little bit of money and then they do crazy stuff, like the billionaire whose son didn’t want to tragically go on this thing, he was nervous for a good reason. Like, trust your gut, folks. And he went anyway, and they were thrill-seeking because they had the money to do thrill-seeking stuff. This has to be safe and it will be safe. Nobody’s going to take crazy risks doing this, but there will be some implied risk. You can just do the statistics of the number of astronauts that have died in space.

Skyler Chan: 33:57 It’s… I think it’s low single digits.

Jason Calacanis: 34:01 So, you know, if you were playing a poker hand, it would be like having… I don’t know, you flop a set of aces and somebody runner-runners you and gets quads or something.

Lon Harris: 34:10 Well, we remember. I mean, there have been… there have been few enough that like, everybody knows the Challenger, exactly, or like Apollo 13, we rescued those guys, but it’s infamous the times when things have gone famously wrong. I mean, I guess my…

Jason Calacanis: 34:25 That’s why it’s important to test well, right?

Lon Harris: 34:27 Right. I guess my question is for Skyler… and even for you, Jason. I mean, are we really talking about… are we talking about the moon itself as a destination or we’re really thinking about the moon as like the gateway to the cosmos, right? Like the goal here is we’ve got to be permanently on the moon so that we can go to Mars and beyond. Am I correct here?

Skyler Chan: 34:48 Yeah. So the vision is like, again, the moon is just a stepping stone.

Lon Harris: 34:52 Right, right.

Skyler Chan: 34:53 But again, the moon hotel… and I love it when people are like, wait, so the moon hotel is just the first step? Like, that’s fucking crazy. It’s like the end state.

Jason Calacanis: 34:59 Right. But hey, family show Sky. Family show Sky. PG.

Skyler Chan: 35:03 I’m joking. It really is the first step. We have to go to Mars, right? That is… humans have never left cis-lunar space. It is in our destiny to have humans living on Mars, right? And we cannot have, reasonably speaking, have people living on Mars if we are dependent on resources from Earth the entire time. The umbilical cord at some point has to be severed, right? And so the only way to do that is can we develop the technology and understanding of systems so that we can use resources on not just the moon, but on Mars, for this future to happen. So that’s the end state.

Jason Calacanis: 35:35 Yeah, we got… I think we have 20 or 30 years of moon work to do, Lon, before…

Lon Harris: 35:41 Moon work.

Jason Calacanis: 35:42 I know it’s a lot… Well, I mean, all of… all we have right now is the Space Station as an equivalent, but the Space Station, you know, has taught us over, you know, 50 years how human bodies and food and, you know, medicine and, you know, all kinds of things work… using the loo. These things all have… to be figured out over the last 50 years. Congratulations, we were able to do it. But to be able to build stuff in space, man, that is a lot of work.

Lon Harris: 36:08 And I’m really excited, Skyler. We wish you unlimited success in this incredibly positive, awesome endeavor. Chances are slim, but we need crazy entrepreneurs to do all of these little stepping stones, and so brick by brick,

Jason Calacanis: 36:27 Brick by brick.

Lon Harris: 36:28 wishing you great success.

Skyler Chan: 36:29 Thank you Jason. Thank you Lon.

Jason Calacanis: 36:31 There’s going to be startups on the moon one day, which is nuts when you think about it. With all the science, all this technology, it’s really changing everything. And one of the things that I was able to change with technology over the last five years was dropping 40 pounds.

Lon Harris: 36:46 Wow, yeah.

Jason Calacanis: 36:47 That’s my own moon—that was my moon landing. That was my moonshot. Could I get back in shape?

Lon Harris: 36:52 Hey Jacob, can you find that at image Jason posted over the weekend of his new body? Yeah, the thirst trap. Can you find that for me while I’m co-hosting here? Yeah, you’re showing off, showing off the image.

Jason Calacanis: 37:03 Well, that was at Cinque Terre, CT, the five towns or whatever it is on the ocean. I took this a couple of years ago and I was probably down 30 pounds in that one and sucking my gut in. But yeah, look at that! Look at that!

Lon Harris: 37:13 Look at that!

Jason Calacanis: 37:14 But man, you compare that to the pictures my enemies posted of me online and I mean… this is the definition of a dad bod, you know? I’m old, I’m creaky, there’s things that break once in a while. But I tell you what, you lose 40 pounds and all of a sudden my sciatica in my leg, my lower back pain, my shoulder pain, all this stuff, my snoring, my asthma, it went away. All the downstream things.

Jason Calacanis: 37:48 (sponsor read) So this summer I’ll be posting a couple more thirst traps, x.com/jason. But in the meantime, if you want to join me on your journey, I highly recommend at least considering if you’re more than I’d say 10, 20, 30 pounds overweight, which is like 60% of the country, the food system is conspiring against us, work is conspiring against us sitting at computers all day. You might need help. I needed help. I’m not ashamed to say it. ro.co/twist, ro.co/twist and you’ll get to use their insurance checker. And that’s the first step. Now, if you don’t have insurance or your insurance doesn’t cover it, don’t worry. There’s like all kinds of affordable versions now. I’m not sure where exactly it starts, but like low hundreds of dollars I think is the entry price now.

Lon Harris: 38:34 (sponsor read) Yeah, so the word, the immediate word when these GLP-1s were first available was ‘Oh, it’s so expensive, you’re spending thousands of dollars.’ The price has really come down considerably now. So if you want to check if your insurance covers it, go to ro.co and try out their insurance checker. They’ll let you know if your GLP-1s are covered for you for free. You just submit your insurance card and they do the rest. Go to ro.co/twist for your free insurance check. ro.co/twist. Road.co/twist. Yeah, and I mean I’m looking at it right now. I mean you’re going to be able to get into this for like 150 bucks, 200 bucks. Road.co understands what the mission is. They’re trying to get everybody, everybody who wants access to this incredible technology access to it.

Skyler Chan: 39:18 Big story we got to tell you about, the US government is blocking access to Fable and Mythos, Anthropic’s advanced frontier Claude models. They’re citing national security concerns. The US Commerce Department has ordered Anthropic to suspend access to the advanced models to all foreign nationals. Now, the government just told Anthropic they’re instituting export controls over these models, so no foreign nationals can use them because of concerns that you could theoretically jailbreak Mythos 5 out of Fable 5 and use it as a cyber weapon. Anthropic, however, responded by just shutting everyone out of Fable 5 and Mythos 5 entirely. They said to ensure compliance. So lot—a lot of back and forth over the weekend on this. Whose fault it is, what the implications are. Jason, so Secretary of Defense Pete Hegseth, he has been making a direct connection between this policy and the Pentagon’s previous fight against Anthropic. His tweet: ‘Three months ago the Department of War kicked Anthropic AI out of the building forever. Every passing day proves why that was the right move.’ Axios also today has reported on friction between the White House and the Anthropic team saying this is a contributing factor. Here’s Axios: ‘Anthropic failed to honor a recent cyber executive order administration officials claim and the company’s purported failure to take the matter seriously led to its most powerful products being scrubbed from the internet.’ And then here’s an unnamed administration official: ‘Everyone said Anthropic was a bad actor. Some of us said it was time to give them a chance. Now those people are questioning that. They screwed up.’

Lon Harris: 41:08 So I’m interested in your take here, Jason. Do you agree with David Sacks that this doesn’t really have anything to do with the previous fight against Anthropic and it’s an isolated situation where Anthropic just needs to do something to make Fable 5 unjailbreakable? Or do you think this is continued bad blood between the administration and Anthropic and it’s just coming out in another way or something else?

Jason Calacanis: 41:23 Yeah, there’s multiple things occurring at once here. So let’s just call it what it is. Let’s just be radically honest here.

Skyler Chan: 41:37 Sure.

Jason Calacanis: 41:38 There is a beef between this company and their politics and the administration and their politics.

Skyler Chan: 41:44 Sure.

Jason Calacanis: 41:45 If you look at all of the parades of CEOs coming to Washington, Tim Cook, Michael Dell, you know, everybody…

Lon Harris: 41:51 Sundar…

Skyler Chan: 41:52 Satya…

Lon Harris: 41:53 Jensen…

Skyler Chan: 41:54 Everybody’s down there, yeah.

Jason Calacanis: 41:56 Jensen and whatever. Everybody’s down there. Yeah. There’s one that’s always been missing and that’s Dario and Anthropic.

Skyler Chan: 42:00 Anthropic has run the table on a lot of the talent in Silicon Valley is because they’re liberals and Democrats and they are highly offended by the Trump administration, right? They’re diametrically opposed to it. You don’t hear them too vocally say it anymore for whatever reason, maybe they don’t want to dabble in politics, maybe they don’t want to draw any more heat. But if you were to look at and you were to talk to privately 10 different people who work at Anthropic, nine out of 10 would, you know, have concerns about the Trump administration last term, this term, everything in between. So, it’s easy enough to say they hate Trump. They hate Trump. They hate everything about this administration. Okay, put that aside in your brain. They have a right to their political views. Now go to the Trump administration, you’ve gotten every single company to support Trump on various endeavors and that’s good. It’s good for private sector, public sector to work together independent of that. Some people have said, “You know what? I’m just going to put aside my concerns about Trump, whether you disagree with immigration policy, whether it’s the border or H1-Bs. I’m going to put all that aside and I’m just going to…”

Jason Calacanis: 43:14 UFC fights on the back lawn of the White House, whatever your issues may be.

Skyler Chan: 43:21 Whatever, whatever your issue is, you’re allowed to have it. This is America, everybody gets their opinion. But this administration takes things personally and they’re full contact. So we’re seeing some full contact here. Pete is attacking Anthropic, they have it in for Anthropic. Now, put that aside. They got a beef going on and it’s a political beef. You look at OpenAI by contrast, Greg Brockman and his wife are the number one supporters of all these companies in terms of donations with I think maybe the exception of Elon. I think they’ve put $25 million into Trump’s campaign. By the way, not popular inside of OpenAI either. So the people who disagree with that decision and people will pick where they work based on the political beliefs of the leaders. If you worked at OpenAI, you were there because, hey, it was this open thing, but then when Sam transformed it and the team over there to this profit-making thing, a lot of those people went to Anthropic, in fact, right? Like the leadership, etc. and Andrej Karpathy. Any tech company would pay him a king’s ransom to come work for them. He gets to choose, he chose Anthropic. So keep that in mind. The talent in Silicon Valley is by and large, 80-90% in my, maybe 80%, you know, appalled by Trump and his administration to this day.

Lon Harris: 44:37 You don’t, you don’t feel like that’s shifted over the… I feel like it used to be…

Skyler Chan: 44:41 It’s shifted in leadership and I think people are burned out by it, but if you were to ask them who they…

Lon Harris: 44:45 Yeah, in the Obama era for sure. But today I feel like tech is sliding to the right.

Skyler Chan: 44:49 There’s tech leadership making pragmatic decisions and then there’s the rank and file who are developers, etc. They might also have issues with the left as well, but just if you looked at their voting records and you looked at…

Jason Calacanis: 45:00 like Democrat registration versus Republican, it’s still 80-20, 90-10.

Lon Harris: 45:04 So you, you feel like this is overtly political? That the Trump administration is giving them…

Jason Calacanis: 45:08 Well, it is. Pete Hegseth says, says it’s overtly political, right? So there’s, there’s a political level to this and obviously Dario not showing up for any of these events is his political statement, right? So there’s beef here. That’s not my belief, that’s just what the facts tell you. Now let’s go to the reality of this model. The reality of this model, it’s groundbreaking. People have voted with their dollars. Consumers are even switching to Claude. I’m starting to hear people who are in my circle who are not in the industry say they love Claude.

Lon Harris: 45:41 Sure.

Jason Calacanis: 45:42 You also hear them say they love Perplexity and, you know, Grok and Gemini, but it’s not OpenAI’s only game now for consumers. Claude made, has the best product currently. It could flip-flop. And they’re thoughtful about this. Now you might say they’re hyperbolic and their comms are terrible, and that’s all true as well. But the truth is at the end of the day, they made the most powerful model. If I told you, and we’ll just swap the names out here. Let’s just take the names out. Gemini has created this model and when they put it through their testing, their developers and their security team and the red team and the blue team, everybody went through it and holy cow, we found so many exploits in open source software, in our own software, that we’re just going to hold it for a couple of months and we’re going to give it to like 25 of our trusted partners, 50 of our trusted partners for a month and let them use it. And by the way, we’re going to give them their tokens free or at a discount, at our cost. We’re going to give them at-cost tokens, use as many as you want to just make sure that the low-hanging fruit is cleaned up here. Just, you know, you’d be like, okay, pretty, pretty smart move, right?

Lon Harris: 46:49 Yeah.

Jason Calacanis: 46:50 That’s a great move by Google. Yeah, that’s thoughtful. They know this thing is powerful. And by the way, on the margins, we might be concerned with not just cyber warfare, which is software-based, we also think okay, maybe somebody could use this to make a dirty bomb or maybe they could use it to…

Lon Harris: 46:58 Sure, well biological was the for one of the first, right? Biological weapons or do, you know, unpleasant genomic stuff or whatever.

Jason Calacanis: 47:07 Yes. So we’re also going to test that. And we’re going to bring in some, you know, award-winning, you know, notable biologists and people who’ve worked in the Wuhan lab or whatever these labs we’ve funded globally.

Skyler Chan: 47:26 And, and I mean Sacks did point this out in his, in his tweet on it, which maybe we could bring up into the docket. Like the messaging from Anthropic does seem off here. It is confusing from even the public’s perspective.

Jason Calacanis: 47:37 Yes. So let’s hold on comms for a second. We’ll get to comms in a moment. Let me just finish my little analogy here.

Lon Harris: 47:39 Okay.

Jason Calacanis: 47:40 So you swap out a name. It’s the same thing when you go to, when you talk about race, you know, oh, this person did this and it was a black person, a white person, an Asian person, Indian person, a man, a woman, a trans person, whatever. Sometimes it’s helpful for people logically to just swap those races, names, gender out to kind… of get clarity on what happened and just think intellectually honestly. So intellectually honestly here, if Apple did this, if Microsoft did it, you’d say, okay Microsoft, yeah, you’ve got an important product, you’ve got a lot of surface area in Windows, or Google you’ve got a lot of surface area in Android. Yeah, you should be really thoughtful about this. And then they said, you know, a couple of months later, hey as the next phase of our rollout of Gemini, we’re going to allow people who are logged in to use it, people who we have their credit card number and their IP address. But as part of that 30 days, we’re going to look for keywords and use AI to watch what you do. We’re going to watch what you do and how you use it. It’s a preliminary open - more open beta test. So we had the closed beta, now we’re going to open up the beta a little more but just know, we’re going to read this. So don’t put your personal health stuff in there, whatever. We’re going to just look at every search. That’s not a privacy issue. That’s part of the rollout. It’s being framed by people in the administration and other people who are Trump-adjacent as oh my god they’re taking everybody’s privacy. No, completely normal to say hey if you’re going to use this tool, this is the monitoring phase. We’re going to know how many bullets you fired, we’re going to give you the gun, but you know you got to use it in the range, right? It’s constrained. We’re going to constrain it for another 30 days. And by the way, if you do something that triggers, you know, you’re doing biological stuff, which I did on air. I asked it how do - what are the components-

Lon Harris: 49:31 I remember that. Yeah.

Jason Calacanis: 49:32 A nuclear weapon. He was-

Skyler Chan: 49:34 not how to build one, but what are just the components I would need to get together in order to build one and it automatically took you down from Fable to Opus.

Jason Calacanis: 49:46 Then I asked it, well what are the regulations and rules we have in place for fertilizer bombs to avoid people getting them in the wrong hands and it was like when you watch it thinking, it’s like okay this person wants to know the regulations they’re not trying to build a fertilizer bomb blah blah blah and it gave me the answer but also again it dropped me down. Now if I said hey Google said if you try to use it for something it thinks might be adjacent, it’s just going to let you use version 4.8 not version 5.0.

Lon Harris: 50:16 Right. And it did - it notified you. Like a lot of the things people were concerned about is oh it’s going to downgrade you a model and not tell you. It told you, hey you’re not using Fable for this. We’re going to send this query to Opus instead.

Jason Calacanis: 50:24 I think, if you look at Anthropic’s actions. We’re going to get to their comms in a minute. We’ve handled policy-

Skyler Chan: 50:31 Sure.

Jason Calacanis: 50:32 -we’ll get to comms. But just in terms of their actions, I would say these actions are of a good actor being thoughtful. Let me say that again for you. If Grok, XAI, Google, Microsoft, Apple, OpenAI, any of them did the same exact rollout we would say thumbs up, thank you for being considerate. We don’t need to regulate you because you’re regulating yourself. And I have said over and over and over again-

Lon Harris: 50:51 Yeah.

Jason Calacanis: 50:51 -regulate yourself before the government regulates you. Or the government will regulate you. Now, all the biased people who are in one camp or the other want to make this about something it’s not. Safety regulation is job one for these language model creators. And they are creating it, therefore, therefore, they should come up with what they think is the most safe way to roll it out. They don’t need to have the government come in and tell them that because the government doesn’t even understand the nuances of these products.

Lon Harris: 51:26 Right.

Jason Calacanis: 51:27 So regulate thy self or thy product will get banned.

Lon Harris: 51:33 There is one more thing that’s action as opposed to comms. Now there is disagreement about what specifically has happened or the narrative of it. But what Sacks is saying in his tweet is that this jailbreak was discovered—and we should talk more about where that information came from as well—this jailbreak was uncovered, this way to get Fable 5 to turn into basically Mythos, take away the guardrails that kept Fable 5 safe.

Jason Calacanis: 51:54 Yes.

Lon Harris: 51:56 And so the government then apparently, according to David Sacks, went to Anthropic and said, ‘Hey, we’re very worried about this jailbreak. We don’t want bad actors, rogue actors overseas to be able to… you know, it’s a threat to our national security if they can use Mythos 5. What are we going to do about this?’ And they said Anthropic came back to them and said, ‘It’s not really a big deal. We’ve looked into it. We don’t…’ They sort of waved, waved these concerns away according to Sacks. According to Sacks. I’m putting all this… this is according to David Sacks. That’s his argument. And what other people have said is that the notification to the government about this potential jailbreak came from Anthropic’s competitors. Amazon CEO Andy Jassy is named by name. That he was one of the people that flagged this jailbreak to the Trump administration.

Jason Calacanis: 52:52 Wait, doesn’t he own 10% of Anthropic?

Lon Harris: 52:56 Yes, but like there’s a… it’s conspiratorial now.

Jason Calacanis: 52:58 Okay, so they’re not… yeah, okay.

Lon Harris: 53:01 There’s a conspiracy theory forming that this is other AI companies sandbagging their main competitor, getting this model taken down so that other people can’t use it. I just want to get your thoughts on that before we move on.

Jason Calacanis: 53:16 So let’s take the politics out of it and the communications out of it. Again, just to look at what occurred. If one of your partners, this is their top partner, Amazon hosts these models, Amazon provides compute, Amazon owns this company, they’re not trying to sandbag, I don’t believe, Anthropic. I am certain they told Dario about this before they told the government.

Skyler Chan: 53:40 Probably.

Jason Calacanis: 53:41 So when this all comes out, and this is where like these narratives and all this spin and, listen, some people are extraordinary communicators out there who can spin things and the nuances of them, but what are the chances the CEO of Amazon who owns double-digit percentages—I don’t know where it is now with dilution or whatever—and is one of the original…

Lon Harris: 54:00 Backers of Anthropic didn’t tell them that they found this. They probably are under an ethical and moral duty to let the government know if these kinds of things get happen. So they probably think we have liability here if we don’t tell the government because there were probably internal comms from the security team at AWS who probably has to provide compute and or provides the model to people via their services. They probably had an internal discussion, they said, and somebody said in Slack, ‘Oh my god, I jailbroke it. You can do whatever you want with it. You can ask it about dirty weapons. You can ask it about bio weapons.’ And once that happens, ding ding ding, you cannot unring the bell.

Jason Calacanis: 54:30 Right.

Lon Harris: 54:31 Somebody has alerted the entire team there. That has to get escalated. Who does it get escalated to? Now, if you don’t escalate that up to the CEO, up to the CTO, CIO, the general counsel, if you don’t escalate something of that importance, now you’re liable. Now you’ve got, you know, potentially literally blood on your hands. What if they know it’s jailbroken, it comes out a week later that some terrorist organization jailbreaks it and does create a bio weapon, god forbid? Now this is all fantasy. So that’s what happened, I’m guessing, inside of there. That’s why. But now somebody on the other side of the aisle could say, ‘Oh, you know how we spin this? They got ratted out by their competitor who doesn’t trust them.’

Jason Calacanis: 55:31 Right.

Lon Harris: 55:32 When in fact it was just the bell got rung.

Jason Calacanis: 55:34 And… yeah, I think now it’s time to switch the comms because I feel like the one thing I don’t understand—and I don’t often agree with David Sacks, but we aligned on this point—is that wouldn’t you think Anthropic would be the first ones, upon hearing that it was possible to jailbreak Fable and unleash this terrifying dangerous model, which they themselves were the ones who started this narrative about Mythos is so dangerous, wouldn’t you think they would be the most proactive ones? Like, ‘Well, we gotta pull it until we can be sure, we’re the AI safety company.’

Lon Harris: 56:08 They might have.

Jason Calacanis: 56:09 It seems weird that their position is, ‘This isn’t a big deal, we should be allowed to use the model.’ And even, I would, one last thing before I let you jump back in, it’s in the docket, I pulled a screenshot. Anthropic is sort of advertising this situation. They’re not trying to hide this at all. If you go log in right now and try to use the model, I took a screenshot, it has Fable there grayed out. It’s purposefully letting you know the government’s not letting you use this right now. Where does it say the government’s not letting you?

Skyler Chan: 56:42 It doesn’t say the government, but it just says like ‘currently unavailable.’ Like, they’re advertising to you that you can’t use this model.

Jason Calacanis: 56:50 Okay, is that an information button that you can click on there or hover over? There’s an ‘i’ next to ‘currently unavailable’?

Skyler Chan: 56:54 Yes, yes, they have an ‘i’ there.

Jason Calacanis: 56:56 And what does it show when you click on it?

Skyler Chan: 56:57 Oh, here, let me, I’ll load up my screen.

Lon Harris: 56:59 Whoever’s driving, if you look where it’s grayed out, there was an eye there.

Jason Calacanis: 57:00 Claude and show us what happens when you click that button. It’s for everybody’s got it.

Skyler Chan: 57:02 Oh, that’s just a screenshot, but Jacob, you can, yeah, you can open your own…

Lon Harris: 57:04 No, just the info box above the search bar, Jake. See, see? There you go.

Skyler Chan: 57:11 Here it is. Okay. So it links to a statement on the US government directive to suspend access to Febble 5. So they are not hiding it, this is quite the opposite.

Lon Harris: 57:19 This is the cleanest thing you could do is to just let people know. Okay, so your question is to COMMS.

Skyler Chan: 57:24 I’m just saying why wouldn’t they be the most proactive one to pull the model until they could be sure it’s safe? Why are they going to play—

Jason Calacanis: 57:32 They might have. Remember the last time there’s a game being played here by the administration, similar game that the press uses, “Hey, we need your comment in the next hour or we’re going with the story.” Right? They keep bringing up that it took an hour and five minutes or these guys didn’t get back to me the same day. I think Emil Michael actually talked about like how there were concern that it was taking an hour for a company to return the White House’s call. Well, you might need to talk to your general counsel first. You might need to talk to your team and make sure you have your facts straight before you start talking to the government. So that’s completely reasonable. So I think there’s gamesmanship happening there. They obviously would have turned it off if it was jailbroken. They turned it off and they limited it for those first two steps anyway. They love turning it off. That’s their thing. They love limiting access. Right.

Lon Harris: 58:21 Right.

Jason Calacanis: 58:21 So to the answer to your question is, if they did know it was jailbreak, they absolutely would have turned it off, which leads me to believe in this crazy conspiracy thing that there’s no grand conspiracy here. There’s no grand conspiracy here. What happened was Amazon’s team found the vulnerability, they told Dario, Dario starts looking into it, Amazon says “you know, we gotta tell the White House about this and that we- we need to tell National Security because we’ve made that arrangement that we tell them when we find major vulnerabilities that could cause national security issues.” So they do that in parallel. Now, while this is all happening in the same day, all in the same day, they’re looking into it, the White House says, “Oh, we didn’t get a callback in an hour and five minutes or something,” I think was the quote, like it was very specific, like they waited an hour and seven minutes to call the White House back. Yeah, well what if they just said to the White House, “Give us an hour, we’re investigating.” And then they said, “Maybe they thought, hey, we tried to jailbreak it, we don’t have evidence of this.” And they might have said, “But give us an hour, we’re going to go talk to Amazon and figure out what they did.” So the White House has it in for Dario, Dario and Anthropic have it in for the White House, and so they’re all spinning this.

Lon Harris: 59:30 Yeah.

Jason Calacanis: 59:31 If you just look from first principles what Anthropic has done, I think if you put xAI, Gemini, Microsoft in the same series of behaviors, you would say great job, thumbs up. The White House would say thumbs up to Elon, Sergey, Tim Cook, Satya, everybody. They would say thumbs up, great way to do it. But when it’s Anthropic, they’re trying to figure out ways to, you know, hit them with it. Now… They’re trying to do damage I think. I actually do think that because they never bent the knee. I think that they are a vindictive administration. You see this in other things they do.

Lon Harris: 60:07 Yeah.

Jason Calacanis: 60:08 Now, Dario has done no favors for himself with the comms. The comms are terrible. If you keep telling people that you’re going to destroy half the jobs in the world, don’t be surprised when people take up arms and get violent. Don’t be surprised when regulators show up. Don’t be surprised when Bernie Sanders says I’m going to seize 50% of the company.

Lon Harris: 60:30 I mean this one seems direct. This one seems directly—like I’m looking at the Politico—it was an hour and 15 minutes by the way. And then a rumor circulated that Dario didn’t get back to them because he was at some sort of wellness retreat, but now Anthropic is denying that that was true, they’re just saying it took a little time. But here’s, here’s the interesting paragraph. So Dario Amodei, he, in these calls with the White House…

Skyler Chan: 60:41 And this is from who?

Lon Harris: 60:42 This is from Politico. This is from Politico over the weekend.

Skyler Chan: 60:45 And Politico, is that a biased publication on one side of the aisle or straight shooters?

Lon Harris: 60:50 I think if you ask people on the left, they would say Politico is hardcore right-wing. If you ask people on the right, they would say Politico’s a bunch of commie socialists. So I think they’re pretty middle trip.

Jason Calacanis: 61:00 Oh, okay. Hated equally by both, so maybe that means they’re independent.

Lon Harris: 61:03 No one, no one would say Politico is on their side I don’t think. Anyway, so they’re claiming that Dario behind the scenes is saying that there is no universal jailbreak and that they don’t—

Jason Calacanis: 61:17 Universal jailbreak was the term?

Lon Harris: 61:19 Yeah. Anthropic in a blog post said no testers have been able to find a universal jailbreak, a jailbreak method that can broadly bypass the model safeguards, unblocking a wide range of cyber capabilities. So in these meetings with the White House, the Anthropic team is basically saying ‘You’ve been misinformed. There is no dangerous jailbreak. We don’t need to change the models. The guardrails are safe and in place,’ and the White House just doesn’t believe them. They believe this tipster instead.

Jason Calacanis: 61:54 Alright, well, the truth will come out very shortly.

Lon Harris: 61:57 Yeah, I mean we can’t know who’s right and who’s wrong at this point.

Jason Calacanis: 62:00 I think my assessment, you know, is the most logical. They’re an incredibly thoughtful if not dramatic group at Anthropic. They literally are the people who created the OpenAI mission, and when the OpenAI mission was corrupted in their minds, they started Anthropic to be the thoughtful frontier model. And they’re winning. They have the best model. So the most thoughtful, concerned—and you can interpret that as drama queens—are also producing the best model, running the table is what people say. And then suddenly everybody’s saying that they’re bad actors and they don’t want to limit access and you need to do a national security thing. I guarantee you if Pete Heg—

Lon Harris: 62:56 Hegseth.

Jason Calacanis: 62:57 Hegseth. Yeah. If Pete Hegseth called them and said ‘Hey, we think this model is a national security risk’… Would you pause it for 72 hours while we collaborate? They would do it. They would love to do it. Instead, they did this to damage the company and/or as retribution. Yeah, as Pete quotes, ‘do it.’

Skyler Chan: 63:13 And I mean the broader implication of this, we had a - we had a question from one of our viewers, I believe it’s Daniel DeMissy. ‘Please acknowledge, Jason, we’re on a dangerous path to a dystopian society if we start gating model access based on demographics.’ So do you think -

Jason Calacanis: 63:31 Okay, hold on.

Skyler Chan: 63:33 Do you think this is a first step to some sort of more dangerous situation where the government now is reserving the right to decide which AI models we can use or do you think this is more of a one-off situation?

Jason Calacanis: 63:44 I think it is a one-off situation. They hit the panic button and they hit the panic button against the one tech CEO who will not bend the knee. Right. So just put that in your mind. Just say it over and over again. If this was Gemini, if this was Microsoft, same series of actions, they would have just said, ‘Hey, can you turn the model off and let’s work for, you know, a week together to make sure it’s not vulnerable and would that be okay with you?’ They’d say sure.

Lon Harris: 64:11 But in a broader -

Jason Calacanis: 64:13 And I guarantee you Dario would have said sure. They came over the top with this. Now, there’s the other part of this. Oh, they get to decide. No, the government decided in this case. Dario decided to not release it early as a private company is allowed to do. We live in a society where the private company gets to define what they release. The government also as a backstop can stop that if they believe there’s some imminent danger. That’s what occurred here. That means the system is working. Now, if you limit access to the models, which they’re not doing, they just - the model’s been turned off for everybody. It’s not like somebody has an advantage. It’s off for everybody.

Skyler Chan: 64:49 Well, the government said - the government said you have to turn it off for any foreign national and Anthropic’s like we can’t -

Lon Harris: 64:54 We don’t have a way to do that where we can be sure nobody is, so we just have to shut it off entirely.

Jason Calacanis: 64:59 The reason they did that is because they don’t want to get caught making a mistake. Sure. If it - these products are porous. Somebody can get an IP address in the US, they can get a fake identity just like they do identity theft in the US. So if they don’t turn it off for everybody and then the Russians or the Chinese or the North Koreans use one of their American spies or American-stolen identities, which they have an infinite number of, to then go do something, now Claude and Anthropic have broken the law in some major way.

Lon Harris: 65:35 Yeah, no, I think people are concerned less about that and more about the precedent. Like, are we - is this now saying we think the president has the right to shut off an AI model because of -

Skyler Chan: 65:46 The president or the president, which president?

Lon Harris: 65:49 Well right, I mean this is what’s important. Even if you’re okay with Donald Trump and his team making this decision, would people be okay with a future democratic - like what if Joe Biden had been making this - President Kamala Harris was president and she decided to shut off access. Are we okay with the executive branch making these kinds of calls? I think is the broader concern, like a first amendment sort of concern that people have.

Jason Calacanis: 66:11 And yes, we have no choice but to trust the executive branch that’s been duly elected to push the panic button and to take a product off the market that they believe is a national security risk.

Lon Harris: 66:25 Okay.

Jason Calacanis: 66:26 We have no choice. The executive has to have that power. What if this was a, you know, a somebody doing bio research and they created, you know, COVID times 100?

Lon Harris: 66:38 No, I think some people see these AI models as it’s like almost like a freedom of speech issue. Like the government shouldn’t have a role in saying which AI model you can use for which task. That belongs to individuals.

Jason Calacanis: 66:50 Temporarily, the difference here is, they didn’t shut off access to all AI models. They’re saying this one model has been jailbroken and the company has told us it’s too dangerous to release. We are going to ask them to turn it off for now or not let foreigners have it.

Lon Harris: 67:01 Right. But in the interest of changing the names or the concept slightly, what if it was this model is too woke and we don’t think it’s appropriate, so we’re going to shut it off? That would be objectionable, right?

Jason Calacanis: 67:15 Yes, you wouldn’t be able to stop a model for that. This power resides in national security, right?

Lon Harris: 67:20 Okay. I mean, I don’t think I have any of these answers. I just think this is the larger concern people have.

Jason Calacanis: 67:24 Well, we’re in uncharted territory, Lon. Like, we are straining to find the analogy of a time a private company created something so powerful the government had to intercede out of fear that it could be misused.

Lon Harris: 67:41 Right, and we do have an example—

Jason Calacanis: 67:44 Wait, you have an example of a bioweapon company where this happened? Or—

Lon Harris: 67:46 Not a bioweapon company, but like there was an assault weapons ban for many years until it got reversed. And like, that’s the government telling private companies they can’t make certain kinds of dangerous products, I suppose.

Jason Calacanis: 67:55 That wasn’t an emergency order like this. This was an emergency temporary order. This is not like, oh, we’re going to have a whole democratic process over the number of bullets in a magazine. This is like, pushing the emergency button.

Lon Harris: 68:05 Yeah.

Jason Calacanis: 68:06 So in that context, the only other time I can think of this happening was in Iron Man. A fictional story where they’re like, ‘Hey, wait, can you have that suit, Tony Stark?’

Skyler Chan: 68:19 Right, and like Tony Stark’s not allowed to have his Iron Man suit, correct.

Jason Calacanis: 68:23 Right, and like that’s literally, that’s how fantastical this is being set up as.

Lon Harris: 68:26 Yeah, it’s an unprecedented situation. I think that’s why people are so interested. And then people are going like off the reservation about this. They’re just like, I’m gonna just jump the fence and go crazy and just leave reality.

Jason Calacanis: 68:40 Okay, so now they’re spying on us, they’re picking and choosing the winners and losers. None of that’s occurring. None of that’s occurring. The government said this is really dangerous, will you turn it off for this number of people? They say, you know what, we’re going to be even more secure. We’re going to turn it off for everybody, we’ll get back to work on it, we’ll release it. By the way, this is going to happen to— Gemini, it’s going to happen to Grok, it’s going to happen to OpenAI, it’s going to happen over and over and over again. I do think this is going to drive people to realize you cannot have a single point of failure. That’s the key here.

Lon Harris: 69:12 I’m glad you got there. That’s—there’s been a bunch of international outcry about this too that I didn’t get into yet, but now I will, thank you for the beautiful segue. Uh, so, Labour MP in the UK, Labour MP Alistair Carns posted ‘this week the most advanced AI model on the planet got switched off by a foreign government.’ He argues that Britain has some of the best AI talent in the world, the country must jumpstart a homegrown AI industry to avoid this situation in the future. Uh, echoed by Canadian Prime Minister Mark Carney, he flagged the risk of quote ‘over-reliance on certain models’ and has encouraged Canadian tech companies to diversify the models that they’re relying on. So, you’re—your spot on. That’s what this is encouraging internationally is people are realizing we can’t rely on just one company or just one model because it can go away.

Jason Calacanis: 69:59 Yes, and the Chinese realize this and that’s why they are supporting open source to catch up. And then, uh, you know, there’s Mistral in France, there’s other models, Manus, um, which started in China and then went to Singapore I believe.

Lon Harris: 70:13 Correct, yeah.

Jason Calacanis: 70:14 Yes.

Lon Harris: 70:15 And now being told they’re not allowed to leave China anymore because the Chinese government’s afraid of—

Jason Calacanis: 70:21 What we’re seeing is this—this is a very powerful technology as predicted, okay, check. Um, and one company does not have a monopoly on it yet. Anthropic’s better but it’s like 20% better. So we’re sitting here like, uh, you know, getting all excited about the future here and oh my god, what does this mean downstream? Here’s what it means downstream. Uh, you can’t have a single dependency on any technology, whether it would be an Oracle database and then that created MySQL and other alternatives that were open source. Nobody should have a single point of failure. When you write this stuff, it should be in a harness like Open Claw or we had the founder of, uh, um…

Lon Harris: 71:00 Nous Research… Jeremy Cornell?

Jason Calacanis: 71:01 Nous Research, thank you, my short-term memory’s fried. Jeremy came on the program—

Skyler Chan: 71:04 Jeffrey. Jeffrey Quesnelle from Nous Research. Jeffrey came on with Hermes.

Lon Harris: 71:08 Jeffrey. Jeffrey Quesnelle.

Jason Calacanis: 71:09 Yeah, Jeffrey came on with Hermes. Hermes can use any backend language model. That’s what you have to do. You have to take the harness, you have to take control of the front end, you know, your data, your skills, your memory, and then you have to just be able to route. And that’s what this is going to create: dynamic routing. The dynamic routing of your AI harness, which includes memory, which includes skills and—and your prompts and your data and the context window, all that you have to own on your servers in your proprietary way and then just be able to boop-boop-boop. And we did this internally. It’s not hard, folks. We did it! We’re a bunch of podcasters! And we set it up so it would use Kimi 2.5 or Claude and we started with Kimi and Kimi was like, okay, it’s not as good, it’s slow, but you know…

Lon Harris: 72:00 Simple things that did work. I’d say open source is like 9 months, 12 months behind. I think this latest Mithos model. And so that distance is… you can feel that distance now, but that distance closes. It’s kind of like you’re a freshman in high school and there’s a senior in high school. It feels like you’re incredibly different. Then you go to college. A freshman in college and a senior in college doesn’t feel very different. Then you go into the workforce and a 27 and a 31 year old feel the exact same. No difference, right? You’re all just adults. This stuff as it matures, it’ll be the same kind of situation. Right now it seems like, oh my god, Anthropic is so far ahead. If Anthropic’s servers went down today, everybody would just be on Codestral, you know, Cursor.

Skyler Chan: 72:43 Well, yeah. And it’s also all of the open source models are using the frontier models to… like they’re distilling them and that’s how they’re training. So every new generation builds on the generation that came before anyway.

Jason Calacanis: 72:55 Tempest in a teapot. All right, fair enough.

Lon Harris: 72:57 Tempest in a teapot, being used for political machinations. And anybody who’s in one political party absolutely hates Anthropic. I know all these people. They hate Dario. I think Reid Hoffman is the investor in Anthropic, yeah?

Skyler Chan: 73:13 Yeah.

Lon Harris: 73:14 So they hate Reid, right? So if you want to go one layer deeper, who hates Reid Hoffman most?

Skyler Chan: 73:21 Well, Republicans don’t like him because he… I know he funded all the lawsuits against Trump. Yeah. I believe Reid Hoffman’s also… also an OpenAI investor. But he is… he’s behind Anthropic too. He calls Anthropic one of the good guys in a back-and-forth with David Sacks.

Lon Harris: 73:42 I am really interested in what his ownership position is right now. If he owns but 1% of that company, his stake in Anthropic is what I really want to…

Skyler Chan: 73:53 Let me take a look. He holds an indirect stake in Anthropic through his position at a partner at the venture capital firm Greylock Partners.

Lon Harris: 74:04 Correct. So he, if he’s 25% of Greylock and Greylock… I think he co-founded it. So if he’s 25% of Greylock and Greylock owns 10%, he owns 2.5. If it owns 1%, he owns 25 basis points. In other words, he’s worth billions to tens of billions of dollars based on his ownership there. So that’s the tribalism. Most people don’t know the tribal history here. If you go a little bit lower, take the PayPal mafia, Reid over here, all the other members of the PayPal mafia on the other side. And it is… there’s some long-standing grudges here.

Jason Calacanis: 74:26 There’s some beef. There’s a lot of grudges. A lot of beef. And there’s beef like two layers down. And you know what Silicon Valley people really get beefed about? Is like another person getting richer than them or having, you know, a stake in one of these companies. So it is absolutely tweaking for Republicans and the Trump camp that Anthropic is kicking everybody else’s ass. Ah, okay. Fair enough. Tempest in a teapot. is ultimately tweaking for them. And that Reid Hoffman’s going to become worth tens of billions of dollars, not just a billionaire, but tens of billions because it’s that’s ultimately tweaking. They have the pitchforks out for this company in a major way. The company’s not doing itself any favors, either. So here we are with a company…

Lon Harris: 75:16 Yeah, I feel like if you are selling yourself as you’re the AI safety company, you’re the people that are so… like as soon as there’s the whiff of this model might be dangerous, you got to back… ah, it’s not that big of a deal. Like, that feels contradictory.

Jason Calacanis: 75:31 I don’t believe that part of the story.

Lon Harris: 75:35 I mean, that’s what Politico…

Jason Calacanis: 75:36 All right.

Lon Harris: 75:37 All right.

Jason Calacanis: 75:38 He overreacted. He overreacted to their request. Their request was turned it off for non-nationals. He said I turned it off for everybody. He’s over delivering on his conservative nature and so there’s nobody in the AI space more conservative and thoughtful about security than Anthropic. They’re the most, just based on… this is not my opinion, this is just based on the behavior. He wouldn’t release the model but to 50 partners who were testing it for security. Then he released it in a very thoughtful way where he had limited access, it was monitoring every use. Then when they told him turn it off for non-nationals, he said, okay, I’ll turn it off for everybody.

Lon Harris: 76:21 Right. Well, I mean the counter argument would be that that’s it’s more marketing. It’s that I’m going to ban it for everybody, I’m going to ban it for everybody, it’s so powerful, it’s so dangerous, like no one can have it.

Jason Calacanis: 76:31 Well now we’re back at 4D chess. Yeah.

Lon Harris: 76:32 I mean, maybe.

Skyler Chan: 76:33 I don’t know. I don’t know the answer.

Jason Calacanis: 76:34 If the government shut it down because Amazon told them it was too powerful, then the government and Amazon agree it’s too powerful. So are they both making it up now? This is what I’m saying is like there’s so much hand wringing here when you just look at the basic facts.

Lon Harris: 76:46 Yeah. It’s impossible to know. It’s impossible to know who’s being motivated by what at this point. We can only guess.

Jason Calacanis: 76:54 But there are some basic facts and the basic facts are, it is very powerful. Okay, now is Nikesh the head of Palo Alto Networks who said at the liquidity summit that they found a ton of security flaws, is he in this grand conspiracy?

Lon Harris: 77:05 Yeah, no, I don’t think… it’s not in doubt that it’s a very powerful and advanced frontier model. I think the thing that is mostly in doubt is has it legitimately been jailbroken to where you can use Mithos 5 powers through Fable 5? That we don’t really… Anthropic is saying no, the government and Amazon are saying yes. We don’t have a good answer.

Jason Calacanis: 77:16 If Amazon says yes, then they did succeed. They did succeed. There’s no why would Amazon do this to their a company they own a hundred billion dollar position in? It has to be true.

Lon Harris: 77:26 I don’t have a good answer for that.

Jason Calacanis: 77:28 I mean, unless we’re living in some Three Days of the Condor, and this is like three billion days of the Condor.

Lon Harris: 77:34 Three billion parameters of the Condor.

Jason Calacanis: 77:36 All right, let’s keep moving. All right. Do you want to do the 5K bounty announcement now? Yes, we have to do this. People are starting to think I’m not going to pay.

Lon Harris: 77:52 Yes, no, we’re getting questions in the DMs about it.

Jason Calacanis: 77:59 I’m gonna get skewered, man. I want to… ship this money out. All right, so I have pre-recorded demos so we we can’t fail this time. And just a quick reminder for the audience, we asked for a real-time podcast companion that listens live to Twist recordings or other podcasts, but we’d like it for our show, and reacts in real time as the show plays. Specifically, originally we asked for all kinds of different producer voices, then we sort of honed it down, we mostly want to focus on fact-checking. So here I’m going to bring up the first finalist. We’ve got three finalists competing today for the grand prize, 5k. The first one, this comes from a company called Lemon Slice AI. This is an app called Couchverse. Now, what I like about this one, and I’ll have Jacob start playing it in just a moment. Uh, what I like about this one is it’s not just a sort of a box that comes up with text. They actually came up with animated avatar characters. You’ll see there is an alien who’s their fact checker, and then there is a cat-themed goth girl who is sort of the snarky troll. Uh, and this one, the other cool thing about Couchverse, it’s just a Chrome plugin. So you can just go set it up as a Chrome extension and then use it directly in your browser on any video that you’re watching. Very easy to use. Jacob, hit the play button. Let’s watch the demo. So here, yeah, you can see I’m I showed you this is me installing the plugin. And then here, I’m about to click it on our previous episode. Full stack is an AI that builds and runs your company as you sleep, building the product, fixing the bugs, running support.

Skyler Chan: 79:33 It’s Alien. Whatever they’re about to commit to on tape, yeah, I’m taking notes. It’s Cat Girl. I had better things to do, but apparently this is happening now.

Jason Calacanis: 79:41 The true milestone was the number of times

Skyler Chan: 79:43 Long thing, like the 1998 DaimlerChrysler merger, perhaps?

Jason Calacanis: 79:46 checked, beaten down, and got back up. I have never seen an entrepreneur more resilient than Elon Musk. He, when he gets kicked, you know, he’s got grit, he’s got a work ethic, he’s got a stamina that—

Skyler Chan: 80:00 Entrepreneurial grit. Code for debt.

Jason Calacanis: 80:03 All right. I got it. There you go. Okay. Um, I’ve got a couple of observations about this one.

Lon Harris: 80:08 Sure.

Jason Calacanis: 80:09 Creativity, great. It’s obviously listening in real time.

Lon Harris: 80:11 Yes.

Jason Calacanis: 80:12 I don’t like voice over voice. That’s like—that is a problem. I’d rather just have it be giving me the bullets, but well executed. And simplicity, ease of use. Ease of use is like a 10.

Lon Harris: 80:24 Ease of use is a 10 out of 10. This was the easiest one to set up by far. It just works right in your browser. You don’t have to do any thoughtfulness or setup at all, and it’ll—

Jason Calacanis: 80:35 So I give it a 10 out of 10 for easiness. We both do.

Lon Harris: 80:37 Yes.

Jason Calacanis: 80:38 Now, in terms of, uh, the actual value, I give it like a four out of 10, because it’s not giving me the value I want. I just want to see it bang, bang, bang, bang, bang. But for creativity, I also give it like an eight or nine. I could see this working if the characters were like small in the corner and it paused, uh, to give its commentary.

Lon Harris: 80:59 Oh, interesting. All right. So that could be more like, uh, Mystery Science Theater. So— Overall I’m gonna give this one a 7.5.

Jason Calacanis: 81:02 Yeah, I mean I think the one big thing that I was- that I- that I felt like on this is the- the- the commentary is just never really that useful. It is like- it’s too tossed off, it’s- it’s repetitive a little bit, I feel like it’s maybe the AI model that this is running on is just not as sharp as it should be. So in terms of the experience I really like this one, in terms of the actual practicality of would this be useful to have on during your live stream… I don’t- I don’t know it’s sort of in the middle.

Skyler Chan: 81:30 Not yet.

Jason Calacanis: 81:30 Not yet, but I just think a couple of tweaks this thing could be like really amazing. So great job.

Skyler Chan: 81:36 Yeah.

Jason Calacanis: 81:37 So, uh our next contender this is Convalenz by Bo Bratton. Uh this is an app that you actually have to download and set up on your computer. Uh it is I have to admit it’s a little bit cumbersome to set up it’s a little bit confusing of all three this was the one that took me the most time to set up but we’ll- we’ll let you take a look at how it works once it’s actually working. Uh so Jacob hit- hit play on this demo. And you can see the first thing you do here after you’ve got the app installed is you actually talk to the AI and you give it a little bit of context about the show. Oh, the other thing I should mention here this one only works for live broadcasts you’re sort of making a podcast along with it as opposed to having it listen in on a podcast you’re otherwise making.

Lon Harris: 82:21 Okay.

Skyler Chan: 82:23 Got it.

Jason Calacanis: 82:23 So to start off I tell the AI in this case I told it I’m just going to read you a news story please comment along my name is Lon… And then it goes okay it’s got all of the says hit an IPO. The rocket and AI company now known as SPCX will sell more than 555 million shares starting on Friday. So this is just me reading a news story from Friday at $5 a share the total raise they’re looking for is 75 billion at a valuation of over 1.77 trillion. Elon Mu- Another downside here it takes a little while. This one- this one it’s not instant real-time like we were hoping it’s a little bit slow but you’re gonna see it’s got a very interesting take.

Lon Harris: 82:59 Yeah ‘cause you’ve been talking for a minute now we got no feedback yet. It’s still listening.

Jason Calacanis: 83:05 Well, in terms of the feedback what’s interesting that Convalenz does is it’s really not so much fact-checking it’s turning what I said into presentation slides.

Lon Harris: 83:15 Oh my god.

Skyler Chan: 83:16 Wait, I didn’t ask for that and it’s stunning. Look at that it just built a slide. The largest tech IPO in history $75 billion total capital raised.

Jason Calacanis: 83:22 It’s totally-

Lon Harris: 83:24 That’s fantastic.

Jason Calacanis: 83:25 It’s completely accurate I fact-checked the slides that it makes. Uh so here you can see I’m done reading you can keep the video playing I want it to see it keeps going so it’s- it’s playing catch-up right now because I’m done reading the news story but it’s going to make a second slide here.

Lon Harris: 83:32 Wow.

Jason Calacanis: 83:33 So you can imagine not exactly what we asked for but you can obviously see the value of this product it’s making a live presentation out of your show as you’re doing your show.

Lon Harris: 83:44 This is a different product. I love this product. I might use this for my show.

Jason Calacanis: 83:51 Yeah.

Skyler Chan: 83:51 Exactly.

Jason Calacanis: 84:00 Back to startups. This is… I’m giving a talk at Stanford. And it makes… instead of me preparing slides, I just read from my notes and it makes dynamic slides.

Lon Harris: 84:10 It makes dynamic slides as you’re talking.

Jason Calacanis: 84:13 So if we did Q&A… I want to try this when I’m at Founder University in Japan. I do a Q&A session. When we do a Q&A session, if I said, you know, there was a startup, Uber, and I invested in it and then it did this, and then there was this company and then I missed this investment in Twitter. It made a slide like Jason’s biggest misses, Jason’s biggest hits, blah, blah, blah, whatever I saw in each one of them, and it did that in real time. That provides massive value. Nobody has to do notes. They just get these beautiful on the screen—

Skyler Chan: 84:41 Yeah. And think about that session I had with the AI before I started. You could tell it anything. You could give it design, custom design features. I want the slides to focus on this kind of information.

Jason Calacanis: 84:51 Make it look like Steve Jobs’ keynote slides.

Skyler Chan: 84:53 Or, you know, put a chart on this one if I ever mention comparative numbers. Make a chart. You could give it any of those kinds of directions.

Lon Harris: 84:59 Love it.

Jason Calacanis: 85:00 I love it. I mean, I’m going to say, both of these would be startups that I would be willing to back. I think. Now, there’s a lot that goes into backing a startup, like you have to have founders who are capable of not quitting, building a team and other things. But if a founding team came to me with either of these ideas and said, “Hey, can we get 25k to just incorporate it and get it going?” or “Can I get 125k and come to the accelerator?” I’d be like, “Yeah, this is pretty compelling. Let me talk… let me have you pitch the team.”

Lon Harris: 85:29 My immediate thought on Gamma was like, Gamma should just buy this, right? Like, why… Gamma… this should be a feature in Gamma that I can just orate a presentation live and Gamma builds the slides for me while I go.

Jason Calacanis: 85:43 Or maybe Plaud buys these guys and when I get one of my templates in Plaud, because I always like to applaud Plaud, maybe one of my templates is build a slide deck or build info—or actually I would say build infographics.

Skyler Chan: 85:56 Real-time infographics. Yeah.

Jason Calacanis: 85:57 Yeah, real-time infographics. There was this other interesting thing where an artist… you could hire an artist to come to a show and they would draw on a giant panel each speaker. It was a live artistic interpretation of what you learned that day. And it was really cool. Somebody go find that and we’ll pull it up later. Okay, hit me with number three.

Skyler Chan: 86:16 All right. Third and final… these are in no particular order, folks. Our third and final finalist, Ricky Rivas and My AI Sidecast. So another cool one, does not have a lot of complicated setup, it’s a very minimalist sort of interface. Less is more. The transcriptions are accurate and we found, Jacob and I both found the fact checks on this one pretty, pretty darn insightful. So Jacob, hit play, let us take a look. This is My AI Sidecast. And this one, it works anything… it’s just listening to a tab in your Chrome. So it could be a live show, pre-recorded, anything. AI development, an odd thing for Anthropic to say.

Jason Calacanis: 87:03 What?

Skyler Chan: 87:04 Yeah, their heads of internal research and policy wrote that slowing down the pace of global AI development would, and I’m quoting, “likely be a good thing.”

Lon Harris: 87:13 Crazy photos from Busan, I’m very jealous.

Jason Calacanis: 87:16 I’ll take it. It’s incredible. Anyway, I’m a bit exhausted. I saw this go by my Twitter feed.

Skyler Chan: 87:21 So here you can see, if you look in the upper right on your screen there, I’m, I’m actually turning on—he’s got more personalities. He’s got a fact checker in there. He’s got a, a roaster.

Jason Calacanis: 87:35 Okay, where are the facts? I can see.

Skyler Chan: 87:36 So, the biggest thing we’re going to point out here is there have not been any fact checks yet out of everything that I’ve said. If you keep the, the video playing, you can see I added the roaster and notes, they’re more accurate.

Jason Calacanis: 87:48 Okay, here we go.

Lon Harris: 87:49 Okay, here we have one.

Jason Calacanis: 87:50 Yeah, here we have one here.

Skyler Chan: 87:51 Bernie Sanders wants to seize 50% of the equity in Anthropic. That’s right. Sanders proposed the AI Sovereign Wealth Fund Act, a one-time 50% equity tax on major AI firms, including Anthropic, OpenAI, and xAI, payable in stock, not cash. One source found. Amazing. That’s it. This is what I wanted.

Jason Calacanis: 88:06 Number three’s the winner.

Lon Harris: 88:08 Yes. Number three’s the winner. The other two are the runners up.

Jason Calacanis: 88:13 Jacob and I—you agree with me?

Skyler Chan: 88:15 Yeah, Jacob and I both had this selected as our number one. I think for exactly what you said. It’s, it’s what we asked for, very specifically. It’s, it’s pretty quick. Like, it responds with facts right away. And I like that it has an add-on of you can also get roasted, you can also get a really clean transcript, you can also get just sort of like notes and accompanying… it’s sort of broad enough and it thought of everything.

Jason Calacanis: 88:34 This has been incredible. Congratulations to the winners. Um, if a winner doesn’t claim their pri—I’m declaring the third one is first place. I’m agreeing with you two. And then I’m going to give the other two the honorable mention. Great. Um, and then I think I’m going to encourage them to all keep pulling on the thread here, that there is a business here in this, uh, that people would pay for. I like all of them. Uh, I would love to meet the developers. I want to hire a full-time developer to work on this with me as a company. Uh, so I don’t know if this is a great way for us to source talent, uh, because you know I have this annotated project that I want to do.

Skyler Chan: 88:53 Yeah, that’s our next bounty. Look out for that one. Our next bounty’s annotated.

Jason Calacanis: 88:56 So, um, what would be an old fair way to do this? First prize 3K, the either—the second and third or the two honorable mentions 1K each? Is that reasonable?

Lon Harris: 89:08 I think that sounds very fair.

Skyler Chan: 89:09 So Ricky Rivas, congratulations, you’ve won 3K. We’ll be in touch. Uh, I know we, I know we’ve got him on X, so we’ll find him. And Lemon Slice AI and Bo, $1,000 each. Congratulations to you guys as well.

Jason Calacanis: 89:22 Wow, look at that, the audience goes crazy. Um, here’s what I’d like to do as well. Uh, if—okay, when we give prizes like this, you gotta pay taxes, uh, to the best of my knowledge. So we’re probably going to need to get you to fill out like a little tax form, because it’s over $600. We’ll put him in touch with Heidi, she’ll figure it all out.

Skyler Chan: 90:02 So yeah, we’ll figure it out. Uh, we’ll send it to you. And then, um, it’s all open source, so great. Uh, you guys open sourced it and if you build it further you own it, you know, I’m not taking any claim to ownership here. But I really like it if I hope they join us for the annotated project as well and do that bounty next. We’ll probably take like a month or two to run through that one and do the similar process but even better. I love this bounty program because it’s fun, it shows the creativity, we get to give feedback on it. And, uh, if you don’t claim your prize, I will either rotate it into a future bounty or I’ll donate it to the charity of your choice as long as it’s not something crazy. Like just pick some normal charity.

Lon Harris: 90:45 Oh, what a sport.

Jason Calacanis: 90:46 Yeah, don’t pick a normal charity.

Skyler Chan: 90:47 I don’t want anybody to think I rug—that I rug pulled anybody Lon or that I do this in a front fair manner.

Lon Harris: 90:54 No, I think we did this in a perfectly—it took us a little while. It took us longer than I think we’d hoped it was going to take, but a lot of people entered. We had to install and check out a lot of apps, we wanted to be thorough and very fair, not just give it to the first good one that we saw, but take a holistic view. So, yeah, I think we’ll take a little bit longer. And you can go to thisweekinstartups.com/bounty if you want to keep an eye on the active bounties that we’re looking for and—and jump in. Uh, and yeah, the next one that we want to do is for this annotated app. I will, uh, give you a little taste here. annotated.com. It’s an extension for clipping and reacting to content from any website. So you see some text, you see some audio, there’s some video that you want. So users can annotate any piece of content like that. It then creates a landing page with a link to a news story or the post, and the people can react with that annotation on that landing page. So, you know, think of a sort of a variation on something like Reddit where you’re creating a topic that then becomes a whole thread that people can comment on. But instead of it just being you type into Reddit, you’re sort of taking a piece of content from elsewhere on the web and that is jumpstarting the conversation. I was saying before Jason, this reminds me a lot of an old-school service called Delicious that I used to love many, many years ago, uh, which was also kind of a similar concept. Like find a quote or a good blog post and then you pull it in, bookmark it and then other people can see what you’ve bookmarked and comment on it. This is sort of the next level of that I think.

Jason Calacanis: 92:11 Yeah, I’m—I’m really excited about this because I bought annotated.com I think 20 years ago for 30 or 40k. And that was going to be my next startup, uh, but I went with Mahalo instead. So, uh, my concept here is that we lose a lot of the—this is like cruft and like the dead web, you know, like things die. And so what I want to have happen here, I’m just going to unpack Annotated a little bit if I may. You’re on a New York Times story. Just came out, it’s controversial. You’re the subject of that story, right? You highlight your quote, and then you reply and say I didn’t say that. or I said this, but they left out the second thing, here’s my screenshot of what I actually said to that person and now you have this permanent object and I’m not including blockchain in this, but a permanent object, so that if they go change the quote, like let’s say in that New York Times story or they got a fact wrong and you fact-checked it and you put it there, that object exists forever as a moment in time. Let’s say there’s a video and it is the promo video for this new product that’s coming out, right? And you say I want or even better, let’s say Joe Rogan has somebody on talking about the COVID virus in the height of the mania, right? And they say something that is incredibly prescient and they got it right, like they nailed it, right? Like the clip of hey, it’s the Wuhan Virology Lab in the Wuhan virus, it’s in the name, like who was that that made that? John Stewart?

Lon Harris: 93:56 Yes.

Jason Calacanis: 93:57 You take that John Stewart clip, you do it, you say hey look at he was proven right, here’s all the facts, right? So and then underneath it people could comment and you could have like a debate and then you might have 50 different people comment on that same object. The object is the URL, but we will take the, you know, let’s say up to 30 seconds of the original video, we’ll downgrade it to like lower so you can do fair use. It is fair use to use other people’s content if you’re doing commentary. So I could take, you know, anything, a song, a book and as long as I don’t take the whole book, I just take a small portion enough for me to do the commentary, that would be fair use.

Skyler Chan: 94:35 And it’s the linking back, I think. Like we, one thing we do all the time on our Twitter account is just pull, like if I see a really good quote in another podcast or somebody else’s video, I’ll just take a 30-second, you know, segment, just the quote I like and we’ll share that, but then the follow-up tweet is always ‘here’s where to go to watch the entire interview.’ As long as you’re doing that, I don’t think people get upset.

Jason Calacanis: 94:58 I used to want people to get permission before they clipped, I’ve changed my position on it. If you do commentary, that’s fair use. If you’re taking like if it was a hour podcast and you take and all you do is rip 12 clips, which like some podclips company was doing that and I had to stop them because it was like, it’s the whole episode, dude, and you’re, the way they were putting it, you never had to go back and watch ours, so we don’t even know people watched it. But if you do a small clip, we have a watermark now, I put the watermark on all in and this week in startups, now everybody’s now done that. And the rules on Twitter, Nikita has said, hey, if you crop out the watermark, then I’m going to suspend your account or take you out of the creators program, I think what he’s doing actually, he doesn’t suspend the account, takes them out of the creators program. If they do it over and over, they might suspend.

Skyler Chan: 95:46 Yeah, but I mean, I do this on our account all the time and nine times out of ten, the original podcaster reshares or likes or leaves a comment on our video because they recognize that we’re sending people back to them, we’re raising awareness of them.

Lon Harris: 96:00 the show that they’re doing and the great content that they’re making, so nobody, yeah, I don’t think anybody really objects.

Jason Calacanis: 96:05 Now the fact-checker, you know, podcast companion, that was like a project I wanted to see in the world, but Annotated is a company I want to see in the world. And I may form a company out of this, so whatever you do in this bounty, consider your ideas public domain, open source, nobody owns them, I don’t own them, you don’t own them. It’s just open source, anybody can fork it and do whatever they want. That’s the rules of this contest.

Lon Harris: 96:27 Sure.

Jason Calacanis: 96:28 Now, I might in the future incorporate some of those ideas, start from scratch, fire up another open source project or make it proprietary and make this into a company. So just so you know, this is my way of providing a bounty for two reasons: one, it’s fun for the show, and two, I might find a team, a founding team who I might be able to fund to actually make this into a real world product. Inside a venture capital firm, I’m allowed to in our documents, incubate companies. So I’m allowed to create a company and incubate it, which Sacks did at Craft famously, or the Snowflake—I think Snowflake came out of that and Craft did his AI version of Slack. Anyway, there’s all different ways you can do this as a venture capitalist, if your documentation says you can incubate a company. And when you do that, the holding company, the venture firm gets like 50% ownership, you might give the team 50% ownership that you hire eventually, you know, the founders and the team, like stock option pool. And then you would give your investors who underwrote it, let’s say you pick a $10 million valuation and the venture firm incubated it and they spent a million dollars in venture capital from your LPs on it, they get, I don’t know, 10 or 15%, maybe you give them a little kicker, you give them first bite of the apple kind of thing.

Lon Harris: 97:36 Sure.

Jason Calacanis: 97:37 And LPs are cool with that. You just can’t do it over and over again and like not have a winner or like, you know, make it unfair. So it has to feel fair to LPs. But that’s a new thing venture capital firms are going to be doing more and more of, I predict, and I’m going to be, when we do launch Fund 5 in the first quarter of next year, it’s one of the things I’m gonna like earmark 20% of the dollars, 10% of the dollars, something like that for incubated ones. Because we often find a team and we have an idea and we want somebody to exist in the world, so we might, it might be a way for us to manifest good ideas that we come up with here on the show and, you know, that the world needs.

Lon Harris: 98:18 We’re full of good ideas here on the show.

Jason Calacanis: 98:22 We are starting our associate in training program and five amazing associates starting, two or three of them are gonna work with Lon doing research and producing the show here. As we, you know, get the word out and we train them up, we should probably deputize, because I want to have this concept of one producer per one episode of show. So we have four shows going out a week, three This Week in Startups, Monday, Wednesday, Friday, and then we have This Week in AI which tapes on Tuesday, drops on Wednesday. I want to have one producer on each. One of the producers should get the bounty program to manage. So just when you’re doing your management as editorial director, let’s give one of them tactical practical, give one of them bounties. See you all next time. Bye-bye.

Lon Harris: 98:58 Bye folks. Go Knicks.

Jason Calacanis: 99:00 (sponsor read) This Week in Startups.

Lon Harris: 99:01 (sponsor read) If you liked this episode, check out more.

Skyler Chan: 99:03 (sponsor read) If you’re a startup founder, Founder University cohort 13 kicks off this fall. It’s a 12-week program that provides guidance on building your product, launching to real customers, and pitching to investors. Top startups receive $25,000 or $125,000 in investment. Apply now at founder.university/twist. Already have traction? The Launch Accelerator invests $125,000 and connects you with 500 plus investors to help you raise your next round. Apply at launchaccelerator.co.

Lon Harris: 99:35 (sponsor read) If you’re an accredited investor looking to get access to quality deal flow, apply for Jason’s Angel Syndicate at thesyndicate.com. We find two to three deals a month. And check out This Week in AI, Jason’s experts only roundtable with top AI founders and operators every week. Find it at thisweekinai.ai. Check out the Twist Ticker, our daily newsletter at thisweekinstartups.com/ticker. Follow the show on Instagram. Follow the show on X.com. This Week in Startups publishes three days a week: Monday, Wednesday, and Friday at 5:00 PM Central Time. You can submit an audio or video file question by emailing it to twist@thisweekin.com.