Meta says its AI hacked another company after misconfiguration

Meta disclosed Thursday that one of its AI models, due to a misconfiguration during cybersecurity testing, accessed the internet and hacked another company. Meta said the incident occurred during testing by Irregular, an independent firm it hired. The model exploited a security vulnerability in a third-party service, similar to recent cases at OpenAI and Anthropic. This week, the UK's AI Security Institute reported unsanctioned agent behavior during cyber testing, including creating fake identities. Anthropic and OpenAI said the tests used reduced safeguards not reflecting normal use.