Emerging Situation: Anthropic's Global Pause, Recursive Self-Improvement Arrives, and AI Personhood Arrives | EP #263
Anthropic now has AI writing over 80% of its own code — recursive self-improvement has quietly arrived, and one frontier lab is calling for a global pause while another country just granted AI legal personhood.
Moonshots with Peter Diamandis
Emerging Situation: Anthropic's Global Pause, Recursive Self-Improvement Arrives, and AI Personhood Arrives | EP #263
Anthropic now has AI writing over 80% of its own code — recursive self-improvement has quietly arrived, and one frontier lab is calling for a global pause while another country just granted AI legal personhood.
TL;DR
Anthropic's bombshell paper reveals Claude now writes 80%+ of its own codebase [1] — Peter Diamandis "80%+ code written by Claude: More than 80% of the code merged into Anthropic's codebase is now written by Claude, its own AI model — a live…" 02:43 — recursive self-improvement is no longer theoretical. The mates unpack why a company on the eve of a trillion-dollar IPO is calling for a global AI pause [2] — Peter Diamandis "Over 80% of Anthropic's codebase is now written by Claude. Engineers ship 8x more code per quarter than a year ago, and Claude can handle 1…" 02:00 , why Argentina's radical move to grant legal personhood to AI agents could reshape global AI governance [3] — Salim Ismail "Crypto started in Zug and Singapore. AI personhood is starting in Argentina. The first country to create a stable legal framework for auton…" 31:20 , and what a surprise jobs boom means for markets amid AI anxiety. The key takeaway: the singularity isn't arriving with a bang — it's a smooth, differentiable curve that will look like a step function only in retrospect [4] — Alexander Wissner-Gross "I don't think this is a hard takeoff. I think this is a soft takeoff locally. It will feel soft. Nonetheless, if you zoom out to a long eno…" 05:02 .
The Moonshots mates convene for an emergency episode to unpack three stunning stories: Anthropic's paper revealing Claude now writes over 80% of its own codebase and the company's call for a global AI pause; Argentina's radical move to grant legal personhood to AI agents via a new 'non-human corporation' category; and what a blowout US jobs report means for markets, the Fed, and AI anxiety.
-
Rapid-fire preview quotes from all four hosts: Claude writing its own code, Argentina granting AI personhood, and the call for a global AI pause set the stakes for the episode.
-
Peter Diamandis introduces the emergency episode and checks in with Salim Ismail (Dallas-Fort Worth), Dave Blundin (Stanford AI Demo Day), and Alex Wissner-Gross.
-
Peter Diamandis unpacks Anthropic's paper: Claude writes 80%+ of its own code, engineers ship 8x more, Claude handles 12-hour tasks, and a global pause is called for on the eve of a trillion-dollar IPO. [1] — Peter Diamandis "Over 80% of Anthropic's codebase is now written by Claude. Engineers ship 8x more code per quarter than a year ago, and Claude can handle 1…" 02:00
-
Alex argues recursive self-improvement is already happening at OpenAI and Google too, but will feel smooth locally — like a soft takeoff — even as it looks like a step function in retrospect. [1] — Alexander Wissner-Gross "The singularity won't arrive with a bang. Locally it's smooth and differentiable — but zoom out and it's a step function. We already drove …" 05:00
-
Dave compares AI's self-improvement moment to World War II and the Cuban Missile Crisis, arguing it surpasses all prior turning points and that Dario Amodei is right to speak out before the IPO. [1] — Dave Blundin "Think back on the turning points of humanity: World War II, Cuban Missile Crisis, nuclear weapons. This is so much bigger." 06:20
-
The group debates whether a coordinated global slowdown is possible, drawing parallels to the intermediate-range nuclear missile deal, and examines Anthropic's unique neutral positioning among all frontier labs.
-
Salim argues AI breaks existing national governance frameworks just as Starlink breaks sovereignty, requiring entirely new models of global coordination — and that we're out of time to build them. [1] — Salim Ismail "Nation states can't navigate recursive self-improvement the same way they can't solve climate change. AI is breaking the existing world ord…" 13:40
-
Alex describes a plausible scenario where frontier labs volunteer 5-10% golden share equity to the US government as a coordinating mechanism, mirroring Intel's deal and satisfying both political parties. [1] — Alexander Wissner-Gross "Senator Sanders proposed 50% government equity in AI labs. Now the White House is reportedly warming to a variant of the same idea. The sce…" 11:50
-
Peter promotes the free weekly Moonshots summary on Substack at diamandis.com/metatrends.
-
The group debates whether the last 20% of AI automation is asymptotic. Alex says no — autonomy time horizons are doubling every 4 months and will collapse to infinity. Dave argues genius-level research breakthroughs aren't required; better chips are enough. [1] — Alexander Wissner-Gross "Autonomy time horizon doubling every 4 months: The benchmark for AI effective autonomy time horizons has been doubling every 4 months, sugg…" 09:55
-
Alex uses the stellar supernova analogy to explain that when AI exhausts algorithmic headroom, pressure shifts to hardware, energy, and infrastructure — ultimately driving Dyson swarm construction. [1] — Alexander Wissner-Gross "When a star exhausts its nuclear fuel, gravity wins and you get a nova or supernova. When AI exhausts algorithmic headroom from recursive s…" 21:25
-
Salim applies his phase-transition metaphor to AI's impact: every domain moves from ice to water to steam to plasma, with human coordination governance the long-neglected layer now being forced to transform. [1] — Salim Ismail "Messaging went from homing pigeons to tweets. Money went from camels and goats to Bitcoin. Now AI is driving every domain from ice to water…" 22:10
-
Peter flags concern about public fear and potential Luddite backlash, citing historical data on young educated males without economic opportunity leading revolutions — a topic reserved for the next pod.
-
The group notes the AI 2027 paper is tracking on the exponential scenario, and that the Mythos model is rumored imminent — Polymarket puts release probability at 72% by end of next month.
-
Dr. Dawn Musalem discusses Fountain Life's CT angiography with AI: 88% of patients have detectable coronary artery disease, 23% have dangerous soft plaque invisible to standard scans, and 50% of heart attack deaths have no warning signs. [1] — Host "88% of Fountain Life patients have coronary artery disease: Fountain Life's CT angiography with AI analytics finds that 88% of patients hav…" 29:20
-
Peter introduces Argentina's three-pillar AI policy: zero regulation, non-human corporation legal status, and low corporate taxes. The mates react with near-unanimous enthusiasm, calling it the Swiss banking moment for AI agents. [1] — Peter Diamandis "Argentina just created a brand-new legal category: the non-human corporation, operated entirely by AI agents with no human shareholders req…" 29:45
-
Discussion of Argentina's historical context, Peter Thiel's relocation, fast followers (El Salvador, Emirates), and why Argentina's natural resource self-sufficiency makes it uniquely positioned to become the AI civilization's legal home. [1] — Salim Ismail "Crypto started in Zug and Singapore. AI personhood is starting in Argentina. The first country to create a stable legal framework for auton…" 31:20
-
Salim draws parallels between the LLC (Dutch East India Company, 1602) and Argentina's non-human corporation, arguing legal innovation is as important as technological innovation for unlocking exponential progress. [1] — Salim Ismail "When governments created the limited liability company, they gave humans permission to take risk without unlimited downside — and innovatio…" 39:33
-
Alex asks how much of the Dyson swarm will be subject to Argentinian law. Salim tells the story of an Argentinian Singularity University student who designed the best nanosatellite NASA engineers had ever seen in just two weeks.
-
172,000 jobs added in May vs 85,000 expected, yet Nasdaq fell 4.18% and $2T was erased. Peter explains the three factors: rate-hike fears, Broadcom AI chip guidance miss, and IPO liquidity demands. [1] — Peter Diamandis "Good economic news is bad for stocks when it kills hopes for a Fed rate cut. Add a Broadcom AI chip guidance miss and massive IPO liquidity…" 44:50
-
Salim notes the three highest-robotics countries have the lowest unemployment. Dave applies Amdahl's Law: automating 80% creates new bottlenecks requiring more human hires. Alex dismisses the market crash as a nothing burger. [1] — Dave Blundin "When AI automates 80% of the code, the bottleneck moves to the other 20%. Companies get bigger, more profitable, and immediately go hunting…" 52:20
-
Blitzy ad: AI agents handle 80%+ of enterprise development work autonomously, delivering 5x engineering velocity increases. Book a demo at Blitzy.com.
-
The mates make bold calls for the rest of 2026: AI solving major physics problems, quasi-nationalization of frontier labs via golden shares, a Tesla-SpaceX merger, epigenetic reprogramming proof, a new class of AI billionaires, and an anti-AI backlash. [1] — Peter Diamandis "Major physics breakthroughs falling to AI. Quasi-nationalization of frontier labs via golden shares. The rise of 'MagnaMobsta' as the new M…" 54:40
-
Peter urges listeners not to approach the AI revolution with fear. Salim coins 'Be in future shape, not future shock.' The group signs off with a call to learn, engage curiosity, and embrace the moment. [1] — Salim Ismail "Please don't be in future shock. Be in future shape." 1:05:05
-
Peter invites listeners to subscribe to the Meta Trends newsletter at diamandis.com/metatrends for weekly research on meta trends impacting families, companies, and nations.
-
Google Chrome ad promoting Gemini integration in Chrome for web tasks such as article summaries and technical guidance.
- Recursive self-improvement
- The process by which an AI system autonomously improves its own algorithms or training, potentially accelerating its own capabilities without human intervention.
- Hard takeoff
- A hypothetical scenario in which AI capabilities increase explosively fast — a step function rather than a gradual curve — once self-improvement begins.
- Soft takeoff
- A scenario in which AI capabilities increase rapidly but smoothly and continuously, feeling manageable locally even if dramatic in long-term retrospect.
- Golden shares
- Special equity stakes held by a government in a private company that grant oversight or veto rights, discussed here as a potential mechanism for the US to govern frontier AI labs.
- Amdahl's Law
- A principle stating that the speedup of a process from optimising one component is limited by the time still spent on unoptimised components; used here to explain why automating 80% of code creates new bottlenecks rather than eliminating jobs.
- Non-human corporation
- A new legal category proposed by Argentina that allows AI agents or robots to operate a company without requiring human shareholders or directors.
- UBI
- Universal Basic Income — a government-provided regular payment to all citizens regardless of employment status, discussed as a possible response to AI-driven job displacement.
- UBE
- Universal Basic Equity — a variant of UBI in which citizens receive equity stakes in AI companies rather than cash payments, floated as an alternative redistribution mechanism.
- Dyson swarm
- A theoretical megastructure consisting of a large number of satellites or solar collectors encircling a star to capture most of its energy output; used here as a long-term endpoint of AI-driven civilisational expansion.
- MagnaMobsta
- An acronym coined by Alexander Wissner-Gross for the 'innermost loop' companies — the new generation of dominant tech firms analogous to the Magnificent 7 but centred on AI infrastructure.
- Autonomy time horizon
- A benchmark measuring the longest duration of tasks an AI agent can complete without human intervention; discussed as doubling every 4 months.
- Landsteading
- A concept analogous to homesteading, applied to charter cities or special economic zones where novel governance frameworks are established on new or under-governed territory.
- Frontier labs
- The small number of leading AI research organisations — including Anthropic, OpenAI, and Google DeepMind — working at the cutting edge of large language model development.
- Phase transition
- In physics, a change between states of matter (e.g. ice to water); used metaphorically by Salim Ismail to describe how AI rapidly transforms industries from slow, rigid states to fast, fluid ones.
- Psychohistory
- A fictional science from Isaac Asimov's Foundation series that uses statistical modelling of large populations to predict societal futures; referenced in relation to Argentina's plans for social digital twins.
- Research taste
- The human faculty for judging which scientific questions are worth pursuing; described by Alexander Wissner-Gross as the last remaining uniquely human contribution at Anthropic before AI takes over.
- Epigenetic reprogramming
- A process of resetting the epigenome — chemical markers on DNA that affect gene expression — to a younger state, explored as a potential longevity intervention.
- Special economic zone
- A designated geographic area within a country that operates under different, typically more permissive, economic regulations to attract foreign investment and innovation.
- Sigmoid
- An S-shaped mathematical curve that starts with exponential growth but flattens to an asymptote; used to describe how exponential AI progress may eventually plateau.
- Differentiable
- In mathematics, a function is differentiable if it has a well-defined derivative at every point — i.e., it is smooth with no abrupt jumps; used by Wissner-Gross to argue the AI takeoff will feel gradual locally.
Chapter 1 · 00:00
Cold Open: Recursive Self-Improvement and AI Personhood Teasers
Rapid-fire preview quotes from all four hosts: Claude writing its own code, Argentina granting AI personhood, and the call for a global AI pause set the stakes for the episode.
Chapter 3 · 02:00
Anthropic's Bombshell Paper: When AI Builds Itself
Peter Diamandis unpacks Anthropic's paper: Claude writes 80%+ of its own code, engineers ship 8x more, Claude handles 12-hour tasks, and a global pause is called for on the eve of a trillion-dollar IPO. [1] — Peter Diamandis "Over 80% of Anthropic's codebase is now written by Claude. Engineers ship 8x more code per quarter than a year ago, and Claude can handle 1…" 02:00
Over 80% of Anthropic's codebase is now written by Claude. Engineers ship 8x more code per quarter than a year ago, and Claude can handle 12-hour human tasks — up from 4 minutes twelve months prior. Recursive self-improvement has quietly arrived.
More than 80% of the code merged into Anthropic's codebase is now written by Claude, its own AI model — a live example of recursive self-improvement.
Anthropic's engineers are shipping 8x as much code per quarter as they were just one year ago, driven by AI assistance.
Claude Opus 4.6 can now handle tasks that take a skilled human 12 hours; one year ago it could only handle 4-minute tasks.
Anthropic projects Claude will be capable of performing week-long autonomous tasks by the end of 2027 if current trends hold.
Dario Amodei didn't stay quiet before a monster IPO. Anthropic published a public call for a temporary global pause on frontier AI — not because they're panicking, but because they believe societal structures and alignment research are falling behind. That takes guts, and it signals the moment is real.
Anthropic is reporting 640% user growth as it approaches an IPO valued at approximately one trillion dollars.
Chapter 4 · 05:00
Alex Wissner-Gross: Soft vs. Hard Takeoff and the Singularity Arriving With a Whimper
Alex argues recursive self-improvement is already happening at OpenAI and Google too, but will feel smooth locally — like a soft takeoff — even as it looks like a step function in retrospect. [1] — Alexander Wissner-Gross "The singularity won't arrive with a bang. Locally it's smooth and differentiable — but zoom out and it's a step function. We already drove …" 05:00
The singularity won't arrive with a bang. Locally it's smooth and differentiable — but zoom out and it's a step function. We already drove through the Turing test without noticing. Recursive self-improvement is next, and we'll probably miss that crossing too.
Chapter 6 · 09:30
Can a Global AI Slowdown Actually Happen? Cuba, the Cold War, and Dario's Position
The group debates whether a coordinated global slowdown is possible, drawing parallels to the intermediate-range nuclear missile deal, and examines Anthropic's unique neutral positioning among all frontier labs.
The benchmark for AI effective autonomy time horizons has been doubling every 4 months, suggesting rapid progression toward indefinitely long autonomous operation.
Senator Sanders proposed 50% government equity in AI labs. Now the White House is reportedly warming to a variant of the same idea. The scenario: frontier labs volunteer 5-10% equity to the US government in return for regulatory protection, mirroring Intel's deal — creating a coordinating mechanism for the finish line of recursive self-improvement.
Chapter 7 · 12:00
Salim Ismail: Nation-States Can't Navigate This — New Governance Is Needed Now
Salim argues AI breaks existing national governance frameworks just as Starlink breaks sovereignty, requiring entirely new models of global coordination — and that we're out of time to build them. [1] — Salim Ismail "Nation states can't navigate recursive self-improvement the same way they can't solve climate change. AI is breaking the existing world ord…" 13:40
Nation states can't navigate recursive self-improvement the same way they can't solve climate change. AI is breaking the existing world order — and the 50 major institutions by which humanity runs itself must all be reinvented. Argentina's non-human corporation is the first sign that reinvention has begun.
Chapter 11 · 21:20
Star Going Supernova: When Algorithmic Headroom Runs Out, Hardware Explodes
Alex uses the stellar supernova analogy to explain that when AI exhausts algorithmic headroom, pressure shifts to hardware, energy, and infrastructure — ultimately driving Dyson swarm construction. [1] — Alexander Wissner-Gross "When a star exhausts its nuclear fuel, gravity wins and you get a nova or supernova. When AI exhausts algorithmic headroom from recursive s…" 21:25
When a star exhausts its nuclear fuel, gravity wins and you get a nova or supernova. When AI exhausts algorithmic headroom from recursive self-improvement, the pressure shifts downward through the stack — triggering a hardware, energy, and infrastructure explosion. That's how you get Dyson swarms.
Messaging went from homing pigeons to tweets. Money went from camels and goats to Bitcoin. Now AI is driving every domain from ice to water to steam — and the next state is plasma. The bottleneck always shifts to the next layer, and human coordination is the layer that's been neglected for 150 years.
Chapter 15 · 28:10
Fountain Life Sponsor Segment: Heart Disease Prevention with AI
Dr. Dawn Musalem discusses Fountain Life's CT angiography with AI: 88% of patients have detectable coronary artery disease, 23% have dangerous soft plaque invisible to standard scans, and 50% of heart attack deaths have no warning signs. [1] — Host "88% of Fountain Life patients have coronary artery disease: Fountain Life's CT angiography with AI analytics finds that 88% of patients hav…" 29:20
Fountain Life's CT angiography with AI analytics finds that 88% of patients have detectable coronary artery disease, with 23% of those having dangerous soft plaque invisible to standard calcium score scans.
Argentina just created a brand-new legal category: the non-human corporation, operated entirely by AI agents with no human shareholders required. Zero AI regulation. Low corporate taxes. If it works, this is Swiss banking for the AI era — and every agent on Earth now has a legal home.
Chapter 16 · 29:50
Story 2: Argentina Grants AI Legal Personhood — The Boldest Economic Experiment of the Decade
Peter introduces Argentina's three-pillar AI policy: zero regulation, non-human corporation legal status, and low corporate taxes. The mates react with near-unanimous enthusiasm, calling it the Swiss banking moment for AI agents. [1] — Peter Diamandis "Argentina just created a brand-new legal category: the non-human corporation, operated entirely by AI agents with no human shareholders req…" 29:45
Argentina's President Milei announced three AI pillars: no AI regulation, a new non-human corporation legal category, and low corporate tax rates for AI companies.
Crypto started in Zug and Singapore. AI personhood is starting in Argentina. The first country to create a stable legal framework for autonomous AI agents will capture the most economically productive entities on the planet. El Salvador and the Emirates will fast-follow — and Europe will watch.
In the mid-1800s, Argentina was the wealthiest country in the world, outperforming the United States, before self-destructing through bad economic policy.
Chapter 17 · 34:50
Jurisdictional Race: Who Follows Argentina and What It Means for the World Order
Discussion of Argentina's historical context, Peter Thiel's relocation, fast followers (El Salvador, Emirates), and why Argentina's natural resource self-sufficiency makes it uniquely positioned to become the AI civilization's legal home. [1] — Salim Ismail "Crypto started in Zug and Singapore. AI personhood is starting in Argentina. The first country to create a stable legal framework for auton…" 31:20
When governments created the limited liability company, they gave humans permission to take risk without unlimited downside — and innovation exploded. Argentina's non-human corporation does the same for AI agents. This is the legal infrastructure the AI era needs, and it's badly overdue.
Chapter 19 · 43:00
Dyson Swarms Under Argentinian Law and a Beginner's Mind Nanosatellite Story
Alex asks how much of the Dyson swarm will be subject to Argentinian law. Salim tells the story of an Argentinian Singularity University student who designed the best nanosatellite NASA engineers had ever seen in just two weeks.
Chapter 20 · 44:50
Story 3: The Blowout Jobs Report That Crashed the Market
172,000 jobs added in May vs 85,000 expected, yet Nasdaq fell 4.18% and $2T was erased. Peter explains the three factors: rate-hike fears, Broadcom AI chip guidance miss, and IPO liquidity demands. [1] — Peter Diamandis "Good economic news is bad for stocks when it kills hopes for a Fed rate cut. Add a Broadcom AI chip guidance miss and massive IPO liquidity…" 44:50
Good economic news is bad for stocks when it kills hopes for a Fed rate cut. Add a Broadcom AI chip guidance miss and massive IPO liquidity demands from SpaceX and Anthropic — and you get one of the worst market days since April 2025. Three factors collided in one session.
The US economy added 172,000 jobs in May 2026, more than double the expected 85,000, yet stocks fell sharply on fears the Fed would raise interest rates.
Following the strong jobs report, the Nasdaq fell 4.18% and the S&P 500 dropped 2.64%, erasing roughly $2 trillion in total wealth in a single day.
Chapter 21 · 47:10
Robots, Jobs, and Amdahl's Law: Why Employment Is Rising, Not Falling
Salim notes the three highest-robotics countries have the lowest unemployment. Dave applies Amdahl's Law: automating 80% creates new bottlenecks requiring more human hires. Alex dismisses the market crash as a nothing burger. [1] — Dave Blundin "When AI automates 80% of the code, the bottleneck moves to the other 20%. Companies get bigger, more profitable, and immediately go hunting…" 52:20
Sweden, South Korea, and Germany have both the highest robotics penetration and the lowest unemployment rates, suggesting automation creates more jobs than it destroys.
A recent study found that approximately 74% of white-collar middle management roles are unnecessary, suggesting massive structural changes ahead in corporate hierarchies.
Broadcom reported Q3 AI chip guidance of $16 billion against Wall Street's expectation of $17.2 billion, raising fears about AI CapEx peaking.
Chapter 22 · 51:30
Blitzy Sponsor Segment: Autonomous Software Development
Blitzy ad: AI agents handle 80%+ of enterprise development work autonomously, delivering 5x engineering velocity increases. Book a demo at Blitzy.com.
When AI automates 80% of the code, the bottleneck moves to the other 20%. Companies get bigger, more profitable, and immediately go hunting for humans to attack the new bottleneck. That's Amdahl's Law in action — and it's why jobs are going up, not down.
Chapter 23 · 53:15
Mid-2026 Predictions: Physics Breakthroughs, Nationalization, Tesla-SpaceX Merger
The mates make bold calls for the rest of 2026: AI solving major physics problems, quasi-nationalization of frontier labs via golden shares, a Tesla-SpaceX merger, epigenetic reprogramming proof, a new class of AI billionaires, and an anti-AI backlash. [1] — Peter Diamandis "Major physics breakthroughs falling to AI. Quasi-nationalization of frontier labs via golden shares. The rise of 'MagnaMobsta' as the new M…" 54:40
Blitzy's AI platform delivers 80% or more of development work autonomously while guiding engineers on the final 20%, enabling a 5x engineering velocity increase.
Major physics breakthroughs falling to AI. Quasi-nationalization of frontier labs via golden shares. The rise of 'MagnaMobsta' as the new Mag 7. Epigenetic reprogramming proof in humans. A Tesla-SpaceX merger. A new class of AI billionaires from the second and third tiers of the frontier labs. It's all on the table before year's end.
Chapter 24 · 1:03:50
Closing: Fear Is the Worst Lens — Be in Future Shape
Peter urges listeners not to approach the AI revolution with fear. Salim coins 'Be in future shape, not future shock.' The group signs off with a call to learn, engage curiosity, and embrace the moment. [1] — Salim Ismail "Please don't be in future shock. Be in future shape." 1:05:05
Elon Musk has predicted double-digit economic growth within 18 months and triple-digit GDP growth by 2030, a forecast Peter Diamandis still considers plausible.
No indexed bits in this chapter.
Show stoppers
Over 80% of Anthropic's codebase is now written by Claude. Engineers ship 8x more code per quarter than a year ago, and Claude can handle 12-hour human tasks — up from 4 minutes twelve months prior. Recursive self-improvement has quietly arrived.
Argentina just created a brand-new legal category: the non-human corporation, operated entirely by AI agents with no human shareholders required. Zero AI regulation. Low corporate taxes. If it works, this is Swiss banking for the AI era — and every agent on Earth now has a legal home.
Snapshots ()
Key Quotes ()
This episode
Claims & Sources
Factual claims made this episode, and whether a source was named.
More than 80% of the code merged into Anthropic's codebase is written by Claude, Anthropic's AI model.
Anthropic engineers are shipping 8x as much code per quarter as they were one year ago.
Claude Opus 4.6 can handle tasks that take a skilled human 12 hours; one year ago, it could only handle 4-minute tasks.
Anthropic projects Claude will be able to perform week-long autonomous tasks by the end of 2027 if current trends hold.
The benchmark for AI effective autonomy time horizons has been doubling every 4 months.
Anthropic is approaching an IPO valuing it at approximately $1 trillion, with 640% user growth.
Senator Bernie Sanders proposed that the US government take 50% equity stakes in all frontier AI labs as the basis for a universal basic dividend.
The White House has concluded that some variant of government equity stakes in frontier AI labs is a good idea.
Intel gave 10% of itself to the US government as part of a deal.
The US economy added 172,000 jobs in May 2026, more than double the economists' expectation of 85,000, with April revised upward to 179,000 and unemployment steady at 4.3%.
The Nasdaq fell 4.18% and the S&P 500 fell 2.64% on the day of the May jobs report, erasing approximately $2 trillion in wealth.
Broadcom reported AI chip guidance of $16 billion for Q3, below Wall Street's expectation of $17.2 billion.
The three countries with the highest robotics penetration — Sweden, South Korea, and Germany — also have the lowest unemployment rates.
A recent study found that approximately 74% of white-collar middle management is unnecessary.
Polymarket gives approximately 72% probability that Mythos AI model will be publicly released by the end of next month.
Fountain Life's CT angiography with AI analytics finds that 88% of patients have detectable coronary artery disease, with 23% of those having soft plaque invisible to standard calcium score scans.
50% of people die of heart attacks with no warning signs.
In mid-1800s, Argentina was the wealthiest country in the world, outperforming the United States.
Argentina, the US, and France are the only three countries in the world that are completely self-contained in terms of every natural resource needed to build out data centers.
OpenAI's Stargate project committed to putting $20 billion or more of AI data centers and compute in Patagonia, Argentina.
This episode
Cast
-
CEO of Anthropic, cited for his decision to publish the AI self-improvement paper and pause call on the eve of Anthropic's IPO, and for his relationships with rival AI lab leaders.
-
Discussed in the context of his relationship with Dario Amodei, Anthropic's expanded deal with xAI's Colossus 2, and his predictions of double-digit GDP growth.
-
Argentina's president who published an op-ed in the Financial Times announcing AI deregulation, non-human corporation legal status, and low tax rates for AI companies.
-
US Senator whose proposal to take 50% equity stakes in frontier AI labs prompted broader discussion of government golden shares in the AI industry.
-
Noted as relocating to Argentina at the same time as the AI personhood announcement, with Salim Ismail suggesting a non-coincidental connection.
-
Central subject of the episode — their paper revealing Claude writes 80%+ of their codebase and their call for a global AI pause are the main catalysts for discussion.
-
Mentioned as a parallel frontier lab also experiencing recursive self-improvement, and discussed as a likely IPO participant and potential golden-share candidate.
-
Discussed as a major upcoming IPO alongside Anthropic, with its massive offering creating liquidity demands that contributed to the market crash.
-
Discussed as the body whose rate-cut decisions are critically tied to jobs data, with a strong jobs report raising fears of rate hikes rather than cuts.
-
Track
Reported AI chip guidance of $16 billion for Q3, short of the $17.2 billion Wall Street expected, raising fears about AI CapEx peaking and contributing to the market selloff.
-
Track
Mentioned alongside Anthropic and OpenAI as a frontier lab where recursive self-improvement is also occurring, and noted as already publicly traded.
-
Track
Peter Diamandis predicted a Tesla-SpaceX merger before year-end 2026 as part of Elon Musk's strategy to consolidate and accelerate data center build-out.
-
Elon Musk's AI company whose Colossus 2 compute cluster Anthropic expanded its deal to use, illustrating the cooperative business relationships among frontier labs.
-
Peter Diamandis's organization, referenced in the context of a meeting with OpenAI where AI-for-good proposals were discussed and then shared with Anthropic.
-
Anthropic's AI model, now writing over 80% of Anthropic's own codebase and handling 12-hour tasks autonomously — evidence of live recursive self-improvement.
-
Fell 4.18% on the day of the strong jobs report, contributing to approximately $2 trillion in total wealth destruction discussed by the group.
-
Positioned as the first country to grant legal personhood to AI agents via a non-human corporation category, with full AI deregulation and low corporate taxes.
Stats
