The results were published on Feb 11 in the Singapore AI Safety Red Teaming Challenge Evaluation Report, meant to flag blind spots to urge developers to fix their models amid growing concerns ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results