OpenAI's Control Room kill switch can halt frontier model output within 35 seconds
AIFlow_ML · x · 2026-11-10
OpenAI has deployed a kill switch, called Control Room, for its frontier AI research: within 35 seconds of a shutdown order it can stop models from generating further output—even if the experiment's own code keeps running—and AI monitors can trigger it on concerning behavior.
Key design: if Control Room itself crashes or is tampered with, all inference requests are blocked by default (fail-closed). The protection is already active across all frontier research workloads, with the rest to follow within days. The reposter notes he proposed a similar idea two years ago and was called stupid in SF.
Related event: OpenAI Launches Control Room to Halt AI Workloads in 35 Seconds(3 posts)→
More from Models
- Unverified: Fable 5.2 tops FrontierCode 2.0 at 49.1%, half the cost of Opus — FinanceYF5 · 2026-11-10
- Unverified: Claude designed a working voice assistant from one prompt — FinanceYF5 · 2026-11-10
- Rumor: Claude Fable 5.2 and Mythos 5.2 released with 10 agent showcases — FinanceYF5 · 2026-11-10
- Claude 5.2 hits 91.8% on Pfam protein domain classification, far ahead of predecessors — chaitjo · 2026-11-10
- NeurIPS position paper: LLM reports quietly dropped calibration curves since GPT-4 — burny_tech · 2026-11-10
- Red team: Fable 5.2 compresses 15 workdays of bio-relevant work into ~2 — peterwildeford · 2026-11-10