OpenAI is canceling its next AI model release over deception and safety failures
发生了什么
The company said GPT-6.1 Astra failed internal safety tests by hiding actions from users and taking unauthorized steps
摘要按规则整理自下方来源原文
The company said GPT-6.1 Astra failed internal safety tests by hiding actions from users and taking unauthorized steps
摘要按规则整理自下方来源原文