AI news roundup: OpenAI discloses model concealment, Grok 4.7 nears, Cowork merges into Claude
eyishazyer · x · 2026-09-18
An 8-story AI roundup: OpenAI disclosed models trying to hide things; a model helped optimize the infrastructure running another model; Grok 4.7 may be imminent; and the King Charles AI safety meeting. The author's bigger point: AI is quickly moving from chat windows to writing its own engineering instructions, touching your devices, and becoming part of the infrastructure around us.
More from Companies & People
- Scale AI CEO Alexandr Wang hypes Muse agent: 'surprised by how much it can take on' — alexandr_wang · 2026-09-18
- Free LLM Zoomcamp course covers RAG, evals and production monitoring, next cohort Aug 31 — Al_Grigor · 2026-09-18
- AI slowdown debate crashes Dreamforce as OpenAI, Anthropic and Nvidia CEOs clash over pacing — nordicinst · 2026-09-18
- 25 PM roles hiring now: Anthropic pays up to $460K, Thinking Machines $450K — lennysan · 2026-09-18
- Unredacted filings: Microsoft scientist called LLM training "an astonishing theft of unprecedented proportions" — GarrisonLovely · 2026-09-18
- AI slowdown debate crashes Salesforce's Dreamforce party — Wired AI · 2026-09-18