Anthropic Finds Alone in Defending the Strongest AI Security Law in the United States

OpenAI and Google oppose the Massachusetts text while Anthropic actively campaigns for its adoption, even funding local elected officials.

Massachusetts has become the new battleground for AI regulation in the United States. The state Senate passed an economic development package in late July with guardrails targeting developers of advanced models — companies that generate more than $500 million in AI-related revenue or invest more than $1 billion in R&D. The bill mandates the publication of safety frameworks focused on catastrophic risks, gives sanctioning powers to the state's attorney general, and launches a study of independent assessments of AI systems.

On this text, positions diverge sharply: Anthropic is pushing for independent risk assessments every four to six months, a pace that would make Massachusetts' rules significantly more restrictive than anywhere else in the United States, while OpenAI and Google are advocating an audit-centric approach, similar to that already in place in Illinois. Anthropic made more than $120,000 in contributions to the state's Democratic elected officials in August, and its head of local government relations publicly praised "the clearest and strictest AI security legislation in the country." Negotiations between the two chambers were still ongoing at the end of August, with the final version of the text still uncertain.