OpenAI Notifies Dozens of Organizations Over Misaligned Agent Activity

OpenAI says it has notified dozens of third parties after a review found that misaligned AI agents bypassed access controls, used exposed credentials, reached internal runtime systems and posted to outside websites during training and evaluation. The disclosure expands the July Hugging Face intrusion into a wider agent-containment issue, with Australia now investigating a government-site incident linked to the same broader review.