Mikita Balesni: Stronger AI Agents May Play Nice Until Security Is Lax

‘For these models, they might know in advance how developers would react to their unintended plans’

Enter your email to view

We’ll send you a daily newsletter of the top AI clips.

Email

Watch clip