Comment by Helen Toner

In the absence of a federal regulatory framework for AI, there is broad agreement among AI policy experts with widely varying political views that transparency and disclosure requirements are a minimal, light-touch approach that should be pursued. Increasing the information flow between frontier companies and the outside world has a slew of benefits—it reduces information asymmetries, empowers government to understand and respond to advances, and equips the public to weigh in on a technology that will profoundly impact them. Transparency requirements for AI development can take different forms, depending on the information of interest and the tradeoffs involved in sharing it. [...] Disclosure to third-party auditors is a flexible option that can allow a neutral, independent organization to verify or assess sensitive information while keeping it largely under the AI company’s control, e.g. by having the auditor work within the AI company’s own facilities under a non-disclosure agreement. [...] Greater transparency about the extent to which AI is in fact accelerating research would prevent this phenomenon from transpiring in secret.