AI progress — 12 November 2026
Nvidia dominates the day’s news: The Verge and TechCrunch report that its board is preparing to remove CEO Jensen Huang, and the stock fell on the reports. Nvidia has not confirmed this, so treat it as developing. Elsewhere, DeepMind’s verification work on Signal and the arrival of Codex on Bedrock are the more concrete items for builders.
New models and papers
- AlphaGenome 2 — Google DeepMind’s updated genetics model, which The Verge says looks at the cellular roots of disease. Why it matters: Researchers working on disease genetics get a new model to evaluate; the coverage does not give benchmark numbers, so the size of the gain is not yet established.
- DeepMind formally verifies key parts of Signal — Ars Technica reports Google DeepMind mathematically verified key parts of Signal’s encryption. Why it matters: It is a concrete case of AI assisting formal verification of security-critical code, which is checkable work rather than a benchmark claim.
- BotPsy — A Hugging Face post describing an on-device personal assistant built on a sub-billion-parameter agentic language model. Why it matters: If it holds up, agent behaviour on phone-class hardware without a server call is something builders can test; the post’s own claims should be checked.
New tools and software
- Gemini Decisions API — Google announces an API for fast, calibrated multimodal decisions. Why it matters: Calibrated outputs matter to anyone wiring model judgments into automated flows; Simon Willison also covered it.
- OpenAI Codex on Amazon Bedrock — AWS documents running OpenAI Codex through Bedrock. Why it matters: Teams already on AWS can use Codex under their existing account and controls rather than a separate vendor relationship.
- AWS emulator — An open repository for emulating AWS, discussed on Hacker News and in an essay by Werner Vogels on the engineer’s job when agents write code. Why it matters: A local emulator gives coding agents somewhere to run and test cloud code without touching a real account.
Overall digest
- Nvidia reportedly preparing to remove Jensen Huang — The Verge and TechCrunch report, citing unconfirmed sources, that Nvidia’s board is preparing to oust its CEO amid allegations about his personal conduct; Nvidia’s stock fell. Why it matters: Anyone buying or planning around Nvidia hardware is exposed to a possible leadership change at the supplier; none of this is confirmed.
- TSMC to invest $1B in Anthropic and OpenAI — Ars Technica reports the chipmaker is investing in both labs. Why it matters: It ties the main chip manufacturer financially to two of its biggest customers’ labs.
- United bans AI agents from booking flights — United will not let AI agents book travel on travelers’ behalf. Why it matters: It is a clear example of a company refusing third-party agents; anyone building a booking agent now has one named service it cannot use that way.
Get the AI digest by email
One email per weekday morning: new models and papers, new tools, and the overall picture. No ads, unsubscribe in one click. See the privacy policy.