OpenAI said that it has built Control Room, a centralised system to track model workloads and stop them quickly when safety or security concerns arise

From the articleHow OpenAI’s Control Room Can Cut Off a Misbehaving AI Model in 35 Seconds - BlockonomiAlso in OpenAI’s new kill switch can stop ‘rogue’ AI models in 35 seconds: How does it work?
Opening sign-in…