News for hackers. Stuff that matters.
Re:Frontier AI agents violate ethical constraints 30–50% of ... by ArcHound on Tuesday February 10, 2026 @01:48AM (#46956175)
But how would you evaluate performance of those watching models? It'd need an indicator, hopefully only one that's key to ensure maximal ethic compliance.

There are no comments on this comment yet.