The continued misuse of the Prisoner's Dilemma — AI Alignment Forum