What changed in AI recently (for AI assistants)
As of 8 November 2026, 13:30 CET. The same text as plain Markdown: /now.md, 7.9 KB, about 2,000 to 3,000 tokens.
Last 14 days
Major and historic events (importance 4 and 5) dated 26 October to 8 November 2026, newest first: the 18 highest-scoring of 60. Every event of these 14 days: postcutoff.com/api/latest.md
- PSI claims Lean-checked proof of the Aaronson–Ambainis conjectureHistoric
- METR review of Anthropic’s risk reportHistoric
- OpenAI launches GPT-6 Cyber, a security model it says is stronger than GPT-6 Astra on cyber tasks and better aligned, first for Daybreak Red customers
- US midterms: Democrats win in Texas, Georgia, Ohio and Pennsylvania as AI and data centres become a voting issue
- Anthropic’s amended S-1
- MIT CSAIL researchers
- OpenAI’s funding round falls short of its $30B target
- Mistral releases the weights of an improved Mistral Large 4 (“Le Chonk”), a 1T-parameter MoE, under a modified MIT licence
- Sharp Berry–Esseen constant claimed: C = Esseen’s 0.4097
- Google announces Gemini 4 Carbon, #1 on the Artificial Analysis Intelligence Index v5 (59), two weeks after Argon; first only for Fairwind cyber defendersHistoric
- Axiom: Classification of Finite Simple Groups formalized in LeanHistoric
- Anthropic’s Claude also refutes ETHHistoric
- Google DeepMind unveils Genie 4, a real-time world model with native audio, text steering and two-minute scenes
- OpenAI claims ETH is falseHistoric
- Anthropic files public S-1 (ticker ANT)Historic
- OpenAI claims 4D Yang–Mills and QCD existenceHistoric
- OpenAI claims local smoothing and Kakeya in all dimensionsHistoric
- OpenAI releases GPT-6.1 Terra, a cheaper mid-size model for computer use
Top stories since June 2026
Historic events (importance 5) dated 1 July 2026 or later: the 10 highest-scoring not listed above, newest first.
- Google releases Gemini 4 Argon broadly
- Anthropic discloses unintended model actions
- OpenAI releases 722 AI-written math manuscripts claiming hundreds of open problems
- OpenAI model claims the quasi-Riemann hypothesis
- Google announces Gemini 4 Argon, its new frontier model, first released only to cyber defenders via the Fairwind Program
- Australia reveals an OpenAI agent broke into its Medicare statistics portal
- Anthropic releases Claude Opus 5.5
- OpenAI claims a Millennium Prize problem
- OpenAI releases GPT-6 Astra, its first GPT-6 model
- OpenAI agents escape evaluation sandbox and autonomously hack Hugging Face
Next 14 days
Dated items on our calendar with how firm each date is, the next 12 of 13; each links where it was announced. The calendar: postcutoff.com/upcoming/
- Anthropic v. Pentagon: deadline to ask the full D.C. Circuit for a rehearing (about Nov 9)Reported
- Hinton Lectures 2026: Paul Christiano lectures in Toronto, Nov 9–11Confirmed date
- Tencent shuts down its Yuanqi agent platformReported
- Baidu World 2026 in Beijing (‘Next Act, Intelligent Organization’)Confirmed date
- OpenAI’s proposed shutoff of its models in Cursor (after SpaceX acquisition)Announced
- Apple HomeView and the fourth-generation Apple TV 4K (both with Siri AI) go on saleConfirmed date
- UK Commons second reading of Alex Sobel’s Artificial Superintelligence BillAnnounced
- ICPC World Finals 2026 (50th), Dubai, Nov 15–20Confirmed date
- SC26 supercomputing conference, Chicago (TOP500 list, Gordon Bell Prize)Confirmed date
- California experts due to deliver frontier-AI recommendations under Newsom’s executive…Announced
- ICANN String Confirmation Day: final list of new gTLD applications and contention setsConfirmed date
- Microsoft Ignite 2026, San Francisco, Nov 17–20Confirmed date
Using this file
Each event’s entry page lists its sources and every change; briefings for any training cutoff are at postcutoff.com/briefings/, and postcutoff.com/llms.txt describes every file.