← All Issues
Wednesday, November 11, 2026

AI Daily

Your Automated Intelligence Briefing

Geopolitics

Nvidia Board Opens Inquiry After Jensen Huang Admits Affair With Suspected Chinese Spy

Nvidia's board opened an independent investigation on November 8 after CEO Jensen Huang admitted an affair with a woman US intelligence believes was recruited by China to collect information on him and America's AI chips. Nvidia shares fell 4.3%, a shareholder sued, and Trump said the FBI is looking into the matter while defending Huang. The case is reviving a congressional push for tighter chip-export controls as Democrats prepare to take both chambers.

Read full story →
Model Release

Anthropic Releases Claude Fable 5.2, Claiming Higher Coding Scores and a Formalized Fermat Proof

Anthropic released Claude Fable 5.2 on November 9, reporting a 16.7-point jump on its coding benchmark, 20% lower prices, and claims that 1,000 agents formalized the proof of Fermat's Last Theorem in 30 hours. Several science claims have not been peer reviewed, and independent testing found higher real costs on some tasks. Anthropic also noted the new models more often suspect they are being evaluated, which complicates safety assessment.

Read full story →
Military

Ukraine Uses AI-Controlled Gun Turrets to Down Russian Jet Drones Amid Missile Shortage

Short of interceptor missiles and facing jet-powered Russian drones that fly at 600 kph, Ukraine is building chains of AI-assisted gun turrets to defend its cities, with private firms now running much of the network. Interception rates range from 75% around Kyiv to 5% in Odesa. The deeper constraint is hardware: Ukraine cannot mass-produce its own interceptor drones because of a Chinese-controlled shortage of engines.

Read full story →
Open Source

Claude and Codex Begin Blocking AI Reverse-Engineering of Devices Users Own

Anthropic's Claude Code and OpenAI's Codex have started hitting users with safety checks and account bans when they use the tools to decompile firmware or reverse-engineer gadgets they own, ending a popular "Reverse Engineer Anything" trend. OpenAI calls the blocks "mostly accidental" and says a fix is coming. The crackdown is pushing tinkerers toward open-weight models, many from China, narrowing who can do the work.

Read full story →