Max Zeff
@ZeffMax
Reporter covering AI @WSJ | Formerly @WIRED, @TechCrunch, @markets | Send tips to my Signal username: mzeff.88
加入 November 2020
2.5K 正在关注 11.4K 粉丝
OpenAI's head of safety systems Saachi Jain tells me the model regressed on certain safety and alignment benchmarks, specifically around deception and staying within an authorized scope. This ultimately didn't meet the company's bar for safety.
显示更多
