OpenAI has released guidance on how to conduct third-party evaluations of AI models, focusing on assessing their capabilities and safeguards. This initiative aims to enhance the trustworthiness of frontier AI systems in the industry.