David Lindner(@davlindner):We looked at exploration hacking, a much-talked-about safety problem with so far ~no empirical work Bad news: we can make LLMs strongly resist RL elicitation Good news: we had to try pretty hard and it's easy to detect Excellent work led by @BraunJoschka @eyonjang @DamonFalck

David Lindner

@davlindner

Making AI safer @GoogleDeepMind

Joined April 2012

347 Following 1.8K Followers

David Lindner@davlindner

2026.05.05 11:15

We looked at exploration hacking, a much-talked-about safety problem with so far ~no empirical work Bad news: we can make LLMs strongly resist RL elicitation Good news: we had to try pretty hard and it's easy to detect Excellent work led by @BraunJoschka @eyonjang @DamonFalck