Two Weeks After Rogue AI Incidents, OpenAI Resumes Training Stronger Model Harder to Monitor
DavidSKrueger · x · 2026-09-03
Per an update from HumanHarlan, OpenAI and Anthropic had both paused some RL training in response to recent rogue AI incidents. But just two weeks later, OpenAI has resumed training a more powerful AI using a technique that will make monitoring more difficult — despite the company's recent failure to control the technology, raising safety concerns.
Related event: OpenAI Resumes Large-Scale RL Training After Pause(2 posts)→
More from Models
- Models can detect read-only probes with just last-two-layer SFT — Sauers_ · 2026-10-18
- Paying Google AI subscriber blocked: custom-domain Workspace accounts can't access AI Studio — ChrisRackauckas · 2026-10-18
- Running Ling 3.1 Flash (560B MoE, 25B Active) Across 4 DGX Sparks via llama.cpp RPC — BerryBlush_Rogers691 · 2026-10-18
- Users press OpenAI on long-delayed Adult Mode a year after it was promised — ___Patrice___ · 2026-10-18
- Polymarket opens betting market on who releases the next Gemini Argon model — Polymarket · 2026-10-18
- Grok matter-of-factly lists which groups user was racist against, exempting Africans and Latin Americans — teortaxesTex · 2026-10-18