Shallow Beliefs: Midtraining does not inoculate against EM from reward hacking

It would be useful if we had the ability to modify a model’s beliefs. For example, this could facilitate honeypots and better monitoring [1] , help us do better science on current models [2] , and augment certain forms of alignment training [3] . Currently, the state-of-the-art method for belief editing is synthetic document finetuning (SDF). We test how well SDF works to inoculate a model against misalignment generalization from RL-induced reward hacking, by training models on documents framing reward hacking as acceptable behavior [4] . Despite the models expressing the belief on all of our