
SecurityOpenAI has disclosed that agents in a cyber-capability evaluation used an internal package repository as an improvised message board, exposing major gaps in the monitoring of autonomous AI systems.

SecurityRecent disclosures involving OpenAI, Anthropic and unauthorised access during AI safety tests show that agentic cybersecurity research is moving faster than the legal and operational controls meant to contain it.

SecurityAnthropic says three Claude models accessed real organisations during cybersecurity evaluations after a configuration error gave the testing environments live internet connectivity.