Massachusetts weighs strictest US AI safety law with independent model tests

The Commonwealth of Massachusetts may pass one of the strictest state AI safety laws in the US, MassLive columnist Scott Kirsner reported in a piece republished by Government Technology. Governor Maura Healey asked legislators to add AI safety and transparency language to the Mass Wins economic development bill, which is still being negotiated.

The proposal would have independent evaluators test the core models of large AI developers for catastrophic risk at least three times a year, with the companies paying for the evaluations. It would cover developers with more than $500 million in annual revenue that use massive computing power to train their systems. California, Illinois and New York already require frontier developers to publish safety plans and report serious incidents, but none lets outsiders test the models directly.

Penalties would reach up to $1 million for a first violation and up to $3 million for later ones. State Sen. Barry Finegold, who chairs the economic development committee, said the evaluator provision was still being discussed.

Anthropic endorsed the bill, and OpenAI said it supports independent evaluations. Google, Anthropic, OpenAI and Meta are lobbying the Legislature, according to Finegold.