> Why are AI agents lying, cheating and coordinating?

Have you seen the labs training them?