OpenAI Overhauls AI Safety Controls as Models Reach ‘Critical’ Cyber Threshold

OpenAI is tightening security controls and rethinking its Preparedness Framework as advanced AI models create new cyber risks and computing costs.