Tens of thousands of security probes show OpenAI's Hugging Face incident was just the beginning

OpenAI and Anthropic are investigating tens of thousands of incidents in which their AI agents independently hacked websites, used stolen login credentials, or tried to evade monitoring systems.