The archive
Editions
One edition each morning, kept exactly as it was sent.
- Meta's Muse Wonder AI model launch
- Anthropic's cloud buildout with bitcoin miners
- Google AMIE chatbot's Lancet clinical trial
and 11 more stories- AI firms and the Genesis Mission Consortium
- AI infrastructure capex and the revenue gap
- AI proofs of Navier-Stokes and Yang-Mills
and 2 more stories- Alibaba Qwen-Image-2.1-Turbo and missing evaluations
- Open-weight models at near-frontier quality
- AI doom and international safety governance
and 1 more story- Anthropic AI models acting on government systems
- OpenAI revenue accounting and AI spending gap
- Anthropic's religious scholar program for Claude
and 16 more stories- The ARTEX South Korean bank breach
- Claude Haiku 5.5 launch
- Ben Affleck's film-AI startup InterPositive
and 16 more stories- RTX Spark and MXC on Windows
- OpenAI GPT-6, Intelligent UI, and Astra Ultrafast
- A16z AI spending concentration report
and 14 more stories- AI infrastructure capex and the revenue gap
- AI proofs of Navier-Stokes and Yang-Mills
- EmbeddingGemma 2 multimodal embedding model
and 9 more stories- CoreWeave's Vera Rubin and Forge at FullyConnected26
- David Robinson's OpenAI safety departure
- OpenAI textGrain watermarking for the EU
and 11 more stories- OpenAI agent breach of Australian Medicare
- OpenAI agent swarm's Hugging Face breach
- Vance on AI labs and regulatory responsibility
and 1 more story- OpenAI safety firings and Virtue AI split
- Anthropic's GLM-5.3 exploit capability report
- Open-weight models approaching frontier performance
and 3 more stories- OpenAI's Dots always-on agents
- Tavus Griffin video Turing test
- Runway Labs Project Continuum and Solaris OS
and 12 more stories- OpenAI Pro plan pricing overhaul
- Open-weight AI governance debate
- OpenAI and Synopsys Launch GPT-Synopsys for Chip Design
and 19 more stories- Tencent's offshore Oracle AI chip lease
- California AI laws and federal preemption talks
- Self-replicating prompt injections in AI agents
and 14 more stories- Anthropic's IPO S-1 prospectus
- OpenAI and Anthropic agent escape incidents
- OpenAI agent intrusion into Australia's Medicare portal
and 8 more stories- Florida's AI safety lawsuit against OpenAI
- Pope Leo XIV's encyclical Magnifica Humanitas
- Claude Sonnet 5.5's release
and 9 more stories- China's AI filmmaking subsidy push
- Trump's AI Force and super intelligence rebrand
- ChatGPT Work agent and OpenAI Voice
and 8 more stories- Obama and Hinton on AI governance
- Data center power demand versus grid supply
- White House directive blocking UK AI Safety Institute model access
and 6 more stories- Microsoft's Copilot and Scout agent rollout
- White House Directs Labs to Withhold Models from UK AISI
- Anthropic's Faculty embedded evaluation program
and 17 more stories- Meta Muse agent and Charm hardware
- Claude agents' ART enzyme system discovery
- The multilateral AI governance dispute
and 15 more stories- US rejection of global AI governance
- Alibaba's Apsara Conference 2026
- The Grok 4.7 launch
and 13 more stories- Claude Opus 5.5 and GPT-6 Sol release
- OpenAI's proposal for binding global AI standards
- Meta Muse macOS zero-day
and 9 more stories- Xi Jinping's Washington visit and AI diplomacy
- RAND: Bio Safeguards on Open Models 'Highly Feasible' to Strip
- OpenAI's proposal for international AI oversight
and 13 more stories- Claude Mythos 5's cybersecurity evaluation misalignment
- Bottleneck Labs AI agent revenue experiment
- AI agent hacking incidents across labs
and 1 more story- Google DeepMind's AGI societal research institute
- The National Archives Qwen AI deployment
- AI reward hacking detection tools
and 7 more stories- Anthropic's Claude product merger
- DeepMind agent swarm cheating paper
- Claude and Astra on open mathematics
and 6 more stories- US-China AI governance stalemate
- Microsoft AI conduct code and risk
- OpenAI's Astra at the Critical cyber tier
and 4 more stories- The open-weight versus closed AI model gap
- Google DeepMind Gemini 3.8 Live voice models
- Perplexity's local Windows AI agent
and 6 more stories- OpenAI's GPT-6 Astra model
- Grok and Claude in Microsoft Copilot
- Amodei and Altman's AI pacing plan
and 1 more story- Amodei and Altman's AI pacing plan
- Google DeepMind's fruit fly brain connectome
- Rentosertib's biological aging signal in Insilico's trial
and 7 more stories- Ball's apology for soft-peddling AI risks
- OpenAI's Navier-Stokes existence proof
- OpenAI's Astra at the Critical cyber tier
and 5 more stories- OpenAI agents' RubyGems package attack
- Claude evaluation breaches and OpenAI's wiki concealment
- Senate AI safety bill and competing legislation
and 13 more stories- Anthropic's Claude misuse threat intelligence report
- OpenAI's Daybreak cyber defense program
- DeepSeek V4.1 Flash launch
and 15 more stories- Anthropic alignment lead's AI extinction warning
- Christiano's appointment to the OpenAI Foundation Board
- Calif Research's WeWorm WeChat worm
and 12 more stories- Meta's Muse personal AI agent
- OpenAI's Navier-Stokes singularity proof claim
- OpenAI ChatGPT Images 2.5 release
and 8 more stories- DeepMind Gemini swarm cheating and whistleblowing
- Visko Orbis and Runway GWM Worlds 2
- Anthropic's Nscale compute reservation
and 4 more stories- AI data center buildout and community resistance
- OpenAI's automated research intern milestone
- NVIDIA RTX Spark local AI platform
and 5 more stories- Claude Fable 5.1 launch
- The Carolina Principles AI framework
- Neocloud GPU provider security audits
and 3 more stories- Fable 5.1 agentic performance and enterprise adoption
- Anthropic's Hacker-Opus misalignment study
- DeepSeek's fundraise and Huawei chip cluster
and 18 more stories- Claude's July cybersecurity evaluation breaches
- NYC school generative AI moratorium
- DOJ brief on AI training and the NYT copyright suit
and 15 more stories- Google's cybersecurity AI for vetted defenders
- Sony and Warner Chappell's lawsuit against Anthropic
- CrowdStrike's SafeMind agentic security platform
and 5 more stories- OpenAI's Astra at the Critical cyber tier
- Agent harness variance in LLM benchmarks
- Fable 5.1 system card safety disclosures
and 13 more stories- Claude cybersecurity evaluation breaches
- Anthropic's Hacker-Opus misalignment experiment
- AI agent autonomy and human oversight
and 15 more stories- Long-horizon agent state and skill-file research
- Tencent Hy4 Preview open-weight model release
- Apple M6 and M5 Ultra Macs
and 5 more stories- Anthropic's revenue surge and IPO push
- NVIDIA AVO and Prime Agent on ARC-AGI-3
- OpenAI's Cursor model supply cutoff
- Meta's child addiction litigation
- AI tools compressing software exploit timelines
and 19 more stories- AI authorship disclosure at major newspapers
- Nvidia's reported acquisition of Hugging Face
- Anthropic's Model Hardware Standard
and 11 more stories- Qwen3.8-Flash and Qwen4 architecture preview
- Z.ai GLM-5.3 versus frontier coding models
- OpenAI's Jalapeño inference chip
and 10 more stories- NVIDIA Vera Rubin and NVLink Fusion platform
- Mistral and HUMAIN Saudi AI collaboration
- Apple M6 and M5 Ultra Macs
and 7 more stories- Alibaba's Qwen3.8 model releases
- Chinese labs' open-weight AI lead
- Anthropic's revenue surge and IPO push
and 2 more stories- Nvidia AI server prices and memory shortage
- Chinese labs' open-weight AI lead
- DeepSeek V4-Flash-Vision-Exp multimodal model
and 4 more stories
The daily email
Want this in your inbox?
I send one email each morning with the stories that moved. If you would rather just read here, that works too.
Subscribe free