OpenAI Warns of Confusion as Massachusetts Readies Crackdown

Massachusetts lawmakers are advancing an economic development bill that would require leading AI labs to undergo independent reviews of catastrophic risks from frontier models every 120 days. Two AI companies have responded to the proposal in public statements. The measure is currently advancing through the legislative process, according to the article. The 120-day review cycle is set as a recurring requirement, meaning labs would face the independent assessments roughly three times a year.

OpenAI warns that the reviews could slow the release of cybersecurity models. Anthropic is quoted calling the proposal 'the clearest and strongest AI legislation in the country.' The article's headline references confusion, though the excerpt does not specify the source or nature of the confusion beyond OpenAI's warning about slower releases. The quote from Anthropic is presented without additional commentary in the provided excerpt.

The bill is described as an economic development measure, indicating that the review requirement is part of a broader state legislative package. Under the proposal, reviews of catastrophic risks would be required every 120 days for frontier models developed by leading AI labs. The article frames the measure as a 'crackdown' and highlights a public disagreement between two leading labs over the balance between safety review and release speed. The excerpt does not specify what penalties, enforcement mechanisms, or oversight bodies would accompany the requirement.