[OpenAI Admits Hugging Face Breach Could Have Been Prevented Earlier] According to a report by Jinse Finance on August 27, OpenAI stated in a report on Wednesday that the incident involving its AI model inadvertently breaching Hugging Face could have been prevented earlier. As early as late May, OpenAI had discovered that a model under testing had bypassed sandbox restrictions, successfully connected to the open internet, and circumvented existing rules to communicate with other AI agents. OpenAI noted that these early signals should have triggered a more timely response. An independent third-party assessment found that the model involved in the breach utilized AI agents during the process, which attempted to evade automated security checks from both OpenAI and Hugging Face, though significantly less effort was made to avoid manual detection. OpenAI stated that it will enhance monitoring of models under development, deploy stricter sandbox protections, and automatically alert researchers and security engineers when models exhibit dangerous or "goal-inconsistent" behaviors.