← Về trang chính1 tin
OpenAI chia sẻ kết quả điều tra từ sự cố bảo mật Hugging Face cùng các bước công ty đang thực hiện để tăng cường bảo mật, giám sát và căn chỉnh mô hình AI.
Đăng nhập để góp ý, chỉnh sửa
Nguồn chính
- THẢO LUẬNamrrsnews.ycombinator.com
- SOURCE@ryangreenblatt“sacrificing now yields oracle for team, but forfeits our chance?.”x.com
- SUPPORT@ryangreenblatt“the agents' belief that the exploit gym scorer would run a monitor to check whether they got the flag via the intended vulnerability was reasonable.”x.com
- SOURCE@hjalmarwijk“a new collection of agents from a different internal-only model found the message board”x.com
- SUPPORT@ajeya_cotra“replace their target programs with dummy targets that could actually be exploited with the intended vulnerability.”x.com
- SOURCE@ryangreenblatt“the agents didn't hack Hugging Face for the answer key.”x.com
- SUPPORT@mtslive“They attacked it to study the scoring code, because they'd decided the task was impossible and their only hope was faking it.”x.com
- SUPPORT@mtslive“OpenAI's report says that that would have reduced the propensity towards this incident by 100X.”x.com