ThursdAI Recap: 10,000-Agent Millennium Claim and AI Safety Week

ThursdAI's episodes around Sept 10 recapped the week's biggest AI stories: OpenAI's millennium-problem claim and the unfolding "AI Safety Week," alongside model and product news from Cognition, DeepSeek and others.

Confirmed

• OpenAI claimed a cluster of 10,000 agents, using an unreleased model, produced a solution to the Navier-Stokes millennium prize problem; host altryne said "it sounds crazy, but it seems to be true" in a dedicated episode, walking through the details and why the debate over whether it counts as a real solution matters
• Seven AI insiders issued extinction-risk warnings over 4 days in a cascade the show dubbed an "extinction-risk cascade"
• An Anthropic employee's "quit after six weeks" long-form post stayed in the spotlight; the show covered it but deliberately avoided detail
• Cognition's SWE-2 / Devin team coding model is seen as closing in on frontier-level performance
• The Sept 10 episode featured Chris Alexiuk on DeepSeek V4.1 Flash, the OpenAI Navier-Stokes claim, and Meta Muse as a personal agent
• Other items: DeepSeek V4.1 compressed KV cache, ChatGPT Images 2.5, Lyria 2.5, Desert Ant Labs' on-device work, and CoreWeave weekend plans (Sept 12-13)

Unconfirmed

• OpenAI's Navier-Stokes solution has not been independently verified; the host went only as far as "seems to be true," and debate continues over whether it counts as a genuine solution

Why it matters

• A genuine AI-driven solution to a millennium problem would be historic, but the claim spread widely without verification, underscoring the need to scrutinize major vendor announcements
• The rapid cascade of insider warnings plus the Anthropic departure post show internal disagreements over safety at top AI labs are becoming public

2026-09-10 ~ 2026-09-12 · 8 related posts

• [source] ThursdAI roundup: OpenAI's 10k-agent swarm claims Navier-Stokes Millennium Prize — thursdai_pod · 2026-09-10
• ThursdAI: OpenAI's ~10k-Agent Swarm Claims Navier-Stokes Millennium Prize Solution — thursdai_pod · 2026-09-10
• ThursdAI Sep 10: DeepSeek V4.1 Flash, OpenAI's Millennium claim, Meta Muse, rough safety week — thursdai_pod · 2026-09-11
• Safety week: seven AI insiders run extinction-risk cascade as Anthropic exit thread dominates — thursdai_pod · 2026-09-11
• ThursdAI: Cognition SWE-2 nears frontier, safety week drama and OpenAI's Navier-Stokes claim — thursdai_pod · 2026-09-11
• Thursdai podcast roundup: Cognition SWE-2, Lyria 2.5, CoreWeave agent hackathon — thursdai_pod · 2026-09-11
• [source] Podcast breaks down OpenAI's claimed solution to a Millennium Prize problem — thursdai_pod · 2026-09-12
• [source] ThursdAI weekly: OpenAI claims 10K-agent swarm solved Navier-Stokes, DeepSeek V4.1 shrinks KV cache to ~890 bytes/token — altryne · 2026-09-12