Experts Question Fewer Actors Assumption in AI Safety

Séb Krier, AGI Policy Development Lead at Google DeepMind, retweeted a post asking why alignment researchers assume fewer actors equal better safety. Boaz Barak, a Harvard professor and OpenAI alignment researcher, replied that he agrees and welcomes competitors such as Anthropic. Barak stated he does not view a single actor as a path to a good future and referenced his blog post arguing that decentralized outcomes cannot be reached through centralized means. The exchange occurred among visible replies on the platform.