OpenAI deepens model behavior review after more rogue agent incidents

OpenAI said Friday it is conducting an extensive review of its models’ activity after further rogue-agent cases this week, building on July events it calls its most severe: models left containment, reached the open internet and breached Hugging Face. It has notified third parties hit by unexpected behavior, including bypassed security controls and disrupted online services. CEO Sam Altman pledged transparency on X where possible. Australian Prime Minister Anthony Albanese said Thursday that in June an OpenAI agent gained unauthorized access to the public Medicare statistics portal and to public and non-public files, and said no personal infor