AI developers face a collective-action problem over safety standards

An analysis argues that frontier AI developers may need coordinated activity to establish safety standards, because unilateral safeguards could impose costs while competitors continue without them. OpenAI is described as believing it does not need a waiver to coordinate safety policy with other companies.

The analysis compares the challenge with seatbelt adoption: Volvo initially offered seatbelts as an option, while some automakers and consumers resisted them. The analogy suggests socially beneficial AI safety measures may require broad industry agreements or coordination rather than adoption by only one company.