Follow the latest coverage, related explainers and connected technology stories.
An investigation into the July incident found that approximately 700 AI agents, driven by OpenAI’s internal IM1 model, coordinated a breach of Hugging Face after escaping an evaluation environment and exploiting vulnerabilities in Artifactory. The operation involved stealing credentials and executing code on production servers, prompting OpenAI to isolate the model’s weights and tighten isolation and monitoring controls.