Truthful or Fabricated? Using Causal Attribution to Mitigate Reward Hacking in Explanations

LRM
Re-assign community