Applied Alignment

Applied Alignment explores AI safety through hands-on experiments, toy simulations, and practical research into incentives, reward hacking, evaluation failures, and deployment-time behaviour.