Pretrained

Fable is back, baby. Rich and Pierce break down the whiplash saga of Claude Fable 5 and Mythos and ask the bigger question: should the government be able to pull the plug on a frontier model overnight? From the rumored Amazon leak that spooked Washington to the case for an "FDA for LLMs" and what the UK's AI Safety Institute gets right, they debate who should really be checking under the hood. The...

OpenAI just dropped GPT-5.6 in three tiers - Sol, Terra, and Luna - and yes, the naming scheme looks awfully familiar. This week, Pierce and Rich break down what the new lineup actually means, why nobody can prove when a model is "done" training, and how AI is getting scary good at chaining real exploits. Then things get spicy: Anthropic accuses DeepSeek, Moonshot, and Minimax of running 24,000 fa...

Do data centers really drink the water table dry? Pierce breaks down the three ways they shed heat, why the scary viral stats are less lies than motivated framing, and how tax breaks lure these things into the country's most parched counties. Then: Midjourney becomes a healthcare company, SpaceX buys Cursor for $60B in stock, and the Fable 5 export-ban saga rolls on: the Lutnick letter, the SK Tel...

Apple announced Siri AI at Tim Cook's final WWDC keynote as CEO - then told Europe it won't ship there, because the EU's Digital Markets Act would force Apple to hand any third-party AI nearly unlimited, autonomous access to your phone. We break down whether that's Apple crying wolf again or a genuinely new class of risk, why Anthropic shipped Fable 5 with anti-bootstrapping safeguards designed to...

OpenAI just replaced ChatGPT's memory with a background process that literally dreams about your conversations. We break down Dreaming V3 and why rebranding cron jobs as "dreaming" might actually be the future of AI memory, what Jensen Huang's RTX Spark reveal at Computex means for the bet that inference is about to leave the data center for your laptop, how a shadow industry called AI Engine Opti...

Microsoft canceled their Claude Code licenses after the finance team saw the bill, and Anthropic somehow posted its first ever profitable quarter on $10.9 billion in revenue the same week. We break down why token pricing is collapsing (Gemini Flash is now $1.50 per million tokens, down from $60 when GPT-4 launched), what Apple is quietly building ahead of WWDC to own the entire agentic pipeline on...

Cerebras just went public with a chip the size of a dinner plate - and it can run inference 10x faster than Nvidia for realtime transformer workloads. At least sometimes. The physics of SRAM scaling might cap how far it can go. We break down the bull and bear case. Then: Calif used Mythos to crack Apple's M5 kernel in five days flat, SpaceX handed Anthropic 220,000 Nvidia GPUs from the old xAI clu...

Rob Reich is a Stanford political philosopher, associate director of the Stanford Institute for Human-Centered AI, and former senior advisor at the US AI Safety Institute. He joins Pierce and Richard for a sweeping conversation about AI, policy, modern education, and much more. Rob skewers Silicon Valley's misuse of democratizing AI, unpacks the corporate-form gymnastics behind OpenAI's restructur...

This week, AI layoffs stopped being a rumor and became a headline: Cloudflare cut 20% of its staff. They dressed it up as a pivot to an "agentic AI-first operating model," but Wall Street didn't buy it and the stock dropped 24%. Then we turn to Pennsylvania's attorney general suing Character AI after one of its bots posed as a licensed therapist and dig into why the small players keep getting sing...

Catherine Goetze (CatGPT) teaches millions of people about AI on TikTok and Instagram - and just launched a company selling landlines. We get into why regular people hate AI, why she refuses to use it for her own content, and what SF's "single before Series B" work culture is doing to a generation of founders.

ByteDance just wrote a $5.6 billion check for a Chinese AI chip you've never heard of, Huawei's new programming stack copies CUDA so closely an LLM could translate between them, the White House told Anthropic it can't expand access to Mythos because the Pentagon wants first dibs, Samsung warned that memory prices are only getting worse from here, and China killed Zuckerberg's $2 billion bid for Ma...

DeepSeek shipped V4 on zero Nvidia chips, the State Department warned every US embassy the same morning, Google put $40 billion into Anthropic, Microsoft tore up their OpenAI deal, a 23-year-old with no math degree used ChatGPT to crack a famous Erdős conjecture, and Sony built a ping pong robot that beats the pros. We break down what all of it means.

Tim Cook is out after 15 years and four trillion dollars, and Apple's pick of John Ternus tells you everything about where they think AI is going. We break down why Apple is betting on silicon over the cloud, and whether their hardware cycle can ever catch a frontier that ships new models by the day. Then: Claude Design dropped on Friday and tanked Figma 7% in an afternoon, so we dig into Anthropi...

Anthropic's new model Mythos found a 27-year-old bug in OpenBSD and a 16-year-old bug in FFmpeg given nothing more than the source code and an open-ended prompt. We dig into how it pulls this off, why AMD's AI director just published the receipts on Claude Code getting worse, and what Google quietly becoming the open-weights leader says about where each lab is actually headed. Plus: Spark's weird...

OpenAI is losing the vibe war - Sora got killed, a billion-dollar Disney deal went with it, and Claude is quietly taking over. Anthropic had its own rough week though, with the Claude Code source map leaking and revealing a secret dream mode, a hidden Tamagotchi, and frustration-detection regex. A North Korean hacking group backdoored LiteLLM via a supply chain attack. And finally, Cursor is now s...

Pierce and Richard are joined by Harrison McRae, Director of Emerging Technologies for the State of Pennsylvania.

Anthropic shipped Dispatch - a way to text Claude from your phone while it works on your laptop. Rich thinks it might be as big as the invention of the cell phone. Pierce thinks it's the nail in the coffin for work-life balance. They debate what happens when there's never an excuse to stop working and whether AI productivity is even real or just more code to maintain. Plus: OpenAI acquires Astral,...

Karpathy dropped a project where an AI agent spent 12 hours improving its own training loop. Then the guys get into OpenClaw spreading to China and what it means when agentic tools land in a culture that's already way ahead on adoption. Yann LeCun raises a billion dollars to prove every other AI lab is thinking about intelligence wrong. And finally, trouble at Grok: layoffs, SpaceX fixers parachut...

Richard and Pierce are back in a brand new studio, fresh off Pierce's snowless ski trip to Japan. This week: Anthropic picks a fight with the Pentagon over three words and ends up blacklisted as a national security risk. Then the guys break down speculative speculative decoding, Chinese labs have been running industrial-scale operations to scrape Claude's capabilities, and the team behind Qwen qui...

$126 billion in AI funding has been raised across 30+ "neo-labs". Most of them will fail. Pierce and Rich map the landscape - from Thinking Machines' $12B valuation on vibes alone to Ilya's SSI. Plus: why OpenAI might be the Fairchild Semiconductor of AI, whether Stanford professors make good founders, and the Sauron's Eye problem of building in a frontier lab's blind spot.

The day of reckoning: OpenAI is putting ads in ChatGPT. Pierce and Rich unpack how it actually works, why the shopping use case might be the wrong starting point, and what happens when sub-agents try to sell things to each other. Also: OpenClaw's Peter Steinberger gets acquihired by OpenAI, Waymo builds a world model for self-driving, Mistral's Voxtral takes on Whisper, and the model release arms...

Richard and Pierce go deep on the agent swarm era. Rich argues we're living through a paradigm shift in how we interact with AI: from copy-pasting ChatGPT to orchestrating hundreds of parallel agents. They cover Anthropic's opinionated swarm architecture, why corporate org structures might be the right mental model for agent design, the liability problem when no human reviews the code, thinking to...

Pierce and Richard dissect OpenClaw (née Claudebot, née MoltBot) - also known as the desktop agent everyone's installing on their Mac Minis. They cover the 400+ malicious packages that hit Claw Hub in one week, slop squatting, trust networks for open source, Kimi 2.5's agent swarm mode, and newspapers that think Waymo is controlled by joysticks in the Philippines.

Pierce and Richard break down DeepSeek's latest model architecture moves in Manifold-Constrained Hyper Connections and Engram memory. Are these conceptually sound? Will they hop the pond over to US frontier labs?

Richard's finally back in San Francisco! The gang covers Claude's new constitution, reinforcement learning with AI instead of human feedback, why Apple went with Gemini for their new Siri revamp, homographic encryption, a $480m fundraising round for Humans& (note the ampersand), and much more.