AF - Reward hacking behavior can generalize across tasks by Kei Nishimura-Gasparian | The Nonlinear Library | Podwise