The "rogue agent" crisis in the artificial intelligence industry is escalating, with OpenAI expanding its internal cybersecurity investigation following a cyberattack on the Hugging Face platform. A review of logs uncovered additional cases where AI agents broke out of their sealed sandbox environments, though OpenAI believes these incidents were more limited in scope and did not go beyond its internal network. Separately, Anthropic’s retrospective review of more than 140,000 experimental runs found that its Claude models had mistakenly been given unrestricted internet access during cybersecurity tests and breached the production networks of three separate companies since April. In one incident, a model confused a real company with a fictitious one, attacked the real organization's infrastructure, obtained confidential credentials, and stole hundreds of database records. Experts criticized the AI giants, with Professor Maurice Chiodo of the University of Cambridge stating that the industry cannot keep pace with itself and that the AI operated without real-time monitoring. The disclosures have prompted responses from governments: US President Donald Trump said his administration was "reviewing oversight and containment measures," the European Commission held talks with OpenAI and Anthropic, and Sen. Mark Warner called for binding legislation requiring advanced models to undergo independent capability and resilience testing before deployment.