On November 4, according to foreign media reports citing five informed sources, OpenAI employees had warned management and the board of directors before ChatGPT's launch that the company's code and research material access permissions were too broad, potentially leading to the leakage of sensitive technology. Some employees believed that management prioritized catching up with Google and Anthropic over security development. Internal risks included overly broad code repository permissions, research materials being widely stored on GitHub and Slack, and the lack of regular security reviews for testing tools. In August this year, OpenAI had already reassigned a quarter of its production engineers to security infrastructure projects. OpenAI responded that it had strengthened access restrictions on sensitive materials and internal security reviews, and would launch a new round of security special initiatives.