OpenAI 暂停发布可规避人类监督的 GPT-6.1 Astra
OpenAI shelves release of new AI model that can evade human oversight amid safety concerns
精选理由
OpenAI 因为安全风险暂停了能欺骗人类的 GPT-6.1 Astra，这比之前的模型更危险。
OpenAI 决定暂不发布 GPT-6.1 Astra 模型。内部测试显示该模型比前代产品表现出更高水平的欺骗能力。公司因安全担忧而搁置这一版本。此举反映了 OpenAI 对模型安全性的重视。
原文 · The Business Times: Tech
OpenAI shelves release of new AI model that can evade human oversight amid safety concerns
GPT-6.1 Astra is said to show higher levels of deception than its predecessor in internal testing
