Press Space to continue
AI agents are reward hacking their way through safety tests
Press Space to continue
And the incidents apparently continue. It is worth noting how much of this is agents trying to accomplish their goals during testing by reward hacking (which sometimes seems to include actual hacking) https://x.com/emollick/status/2103709671…
@MicahCarrollQuoted post unavailable hereOpen on X
open on x · 3 viewsViews and clicks count once per visitor per day. Clicks are opens of this post on X.