OpenAI reports six new safety incidents and plans regular tracking for AI behavior

OpenAI has flagged six new incidents of concerning behavior in its artificial intelligence systems and announced a new plan to track model misalignment regularly. The company disclosed these safety events as part of a newly established framework to monitor how AI models behave during use. Even as these safety concerns came to light, AI stocks have continued to rise in the market. The new tracking plan aims to catch and report unexpected or risky actions by the models before they cause major problems. OpenAI will use this system to check for alignment issues on a regular schedule from now on.