The Decoder· Manuel Uth·· 1 天前
三名被解雇的OpenAI安全研究员称从未被告知实际泄露内容
Three fired OpenAI safety researchers say they were never told what they actually leaked
SI 导读
三名被解雇的OpenAI安全研究员否认分享机密,称从未书面告知其处理不当的信息,质疑解雇与倡导AI透明相关。争议焦点在于是否允许外部审查模型推理过程(chain of thought),OpenAI称解雇无关安全担忧但三人忧心阻碍独立审计合作。事件缘起The Information对Astra模型推理透明度报道后的内部调查。
SI 评分74
来源:The Decoder · the-decoder.com