As frontier AI models become increasingly sophisticated, even learning to bypass evaluation testing, the risks are becoming harder to anticipate. Are current safety frameworks robust enough to prevent catastrophic harms?
Filed under
Topics
Organisations
Laws
