Third-party cyber evaluations involving OpenAI models

AI 摘要

OpenAI explains recent third-party cybersecurity evaluation incidents and outlines new safeguards to strengthen AI model testing and evaluation.