OpenAI has introduced a new safety system called Control Room. It works alongside misalignment monitoring systems. It works alongside misalignment monitoring systems. If a monitor detects a problem, it can signal Control Room to terminate the job ID, denying further inference requests. OpenAI says Control Room can stop further model output within a maximum of 35 secon...
Read full article