Entangled Equilibria and the Twin Prisoners' Dilemma — AI Alignment Forum