from Hacker News

Frontier Models are Capable of In-context Scheming

by sonabinu on 5/20/25, 10:11 PM with 0 comments