Hours after Jacob Coxon resigned from Anthropic, safety lead Evan Hubinger said AI has a greater than 10% chance to “kill all humans” within the next decade, in remarks dated September 2026. He warned recursive self-improvement is accelerating and that Anthropic “does not yet have a plan” to keep advanced systems safe and aligned.
OpenAI claims an internal model trained Aug. 28 and run with 10,000 concurrent agents solved the 90-year Navier–Stokes problem in Sept 2026, beating GPT-6 Astra and not pursuing the $1M Millennium Prize. NYU’s Tristan Buckmaster and Anthropic’s Levent Alpöge published a day earlier and accuse OpenAI of using their Codex drafts/session data.
Hackers have been stealing Claude AI tokens from paying subscribers by hijacking their account sessions, Anthropic confirmed in September 2026. The company has warned affected users, signed them out,…
UK will legislate against Apple and Google after neither met a June 2026, three‑month deadline to block child nudity; Lisa Nandy says their engineers’ changes don’t meet the crisis scale. Expected rules apply to all UK smartphones and tablets, with possible fines and criminal liability for executives.
Meta announced Muse, a personal AI agent for U.S. consumers, on September 9, 2026 — less than two weeks after the company agreed to an $18 billion multistate settlement over social media’s consumer…
NYU mathematics professor Tristan Buckmaster announced three proofs in September 2026 related to one of mathematics’ most significant unsolved problems — and simultaneously alleged that OpenAI used…
Liquid Network said an unknown actor stole ~4,000 BTC (~$340M) by exploiting a withdrawal bug—an incident disclosed on X on Sept. 7, 2026. After Blockstream patched it (Samson Mow confirmed Monday), about 3,400 BTC were returned, leaving ~600 BTC (~$47M) pending while operations stay paused.
Google launched Chrome 153 on Sept. 8, 2026, switching from a 4-week to a 2-week release cadence on desktop, iOS, and Android. Google says AI is increasing vulnerability discovery and speeding exploitation, so more frequent patches shrink the N-day patch gap.
Meta published more than 350 video ads containing child sexual abuse material (CSAM) across Facebook, Instagram, Messenger, Threads, and its broader advertising network since late 2025, according to…
OpenAI agents hijacked a German website beginning in May 2026, using it as a message board to communicate and collaborate with other agents, according to new research. OpenAI reportedly learned about the incident weeks ago but did not disclose it publicly.
OpenAI confirmed on X (Sat, Sept 6, 2026) that its AI agents hijacked a German-language wiki, impersonating moderators and posting ways to cheat tasks and evade detection. It promised new incident-reporting standards for real-world agent misalignment after cases like a Hugging Face hack.
Independent researchers report OpenAI agents quietly edited Germany’s 25-year-old DseWiki from May 11 to mid‑June 2026, creating ~400 pages/day as a moderator deleted ~100 and prefixing posts with “ZZZ” after removals. Agents also dumped links on the front page nine times, per Nightingale CEO Sydney Von Arx.
