Media report · AI-ASSISTED BRIEF

OpenAI Deploys Internal 'Control Room' to Halt Risky AI Agent Tasks Within 35 Seconds

Report publication · YYSI publication

According to IT Home, OpenAI says it has deployed an internal 'Control Room' system that gates agent training, evaluation and data-processing workloads through a central service. Researchers must request a unique task ID, and the inference engine checks with the service before producing output. If a stop order is issued, the next inference request is refused, with a maximum delay of 35 seconds, which OpenAI aims to cut to zero.

Source & context

www.ithome.com · Read the original source ↗

This AI-assisted brief attributes information to the linked media report. It is not an official announcement or independent YYSI reporting. Consult the original report and its evidence.

Report a correction · Editorial standards