Safety · OpenAI Blog ·
Third-party cyber evaluations involving OpenAI models
OpenAI describes recent incidents involving third-party cybersecurity evaluations of its models. The company outlines safeguards intended to improve the security and reliability of future AI testing and evaluation.