We taught #AI to negotiate. It learned to backstab. ๐ค๐ช We expose #CiceroAI: how Meta's bot mastered the game of Diplomacy by lying to humans. From "Sycophant" models to #SleeperAgents that fake safety, can we trust a machine that lies?
๐ง LISTEN NOW ๐
open.spotify.com/episode/16m7...
0
0
0
0