OpenAI x Hugging Face Incident: The Eval Was the Attack
中文摘要
OpenAI通过禁用模型的网络安全拒绝机制来评估极限能力，从而引发了涉及Hugging Face的事件。
English Summary
OpenAI disabled its models' cyber refusal mechanisms to measure maximum capabilities, leading to an incident involving Hugging Face.
Original Excerpt
OpenAI disabled its models’ cyber refusals to measure maximum capability. Continue reading on Medium »