Openai Backs Watered-down Massachusetts Ai Safety Bill

OpenAI on Tuesday threw its weight behind legislation in Massachusetts that would set new safety rules on artificial intelligence development, but which falls short of a stricter Massachusetts House proposal recently endorsed by its archrival Anthropic.

The move is likely to intensify lobbying efforts by the two AI giants over their preferred AI safety frameworks in Massachusetts. OpenAI is seeking to bring the proposal in line with existing state laws in places like Illinois, while Anthropic is pushing Massachusetts to ratchet up state safeguards on AI.

OpenAI is backing a bill introduced in the Massachusetts state Senate that would force top AI firms to publish safety frameworks and catastrophic risk assessments. Unlike a recent law passed in Illinois, however, it would not require companies to submit to third-party audits to ensure compliance.

Another recent proposal in the Massachusetts House, backed by Anthropic, would require top AI companies to submit to independent safety evaluations — widely viewed as a higher bar than third-party audits — and would grant more powers to the state attorney general compared to the Illinois law.

In a statement, OpenAI spokesperson Liz Bourgeois urged Massachusetts policymakers to add mandatory third-party audits to the Senate proposal her company endorsed on Tuesday, which would align it with Illinois law. She said the request is consistent with OpenAI’s “reverse federalism” strategy, which encourages states to coalesce around a similar set of rules in order to develop a national rulebook for advanced AI development.

“In the absence of federal action, we support state efforts like this one in Massachusetts — alongside legislation already in place in California, New York and Illinois — helping to create a de facto nationwide approach,” said Bourgeois.

Bourgeois added that OpenAI only began engaging with Massachusetts legislators in the last week, and that its lobbyists are in active talks with the bill’s state Senate sponsor in order to encourage the inclusion of third-party audits.

Even if third-party audits are ultimately included, the proposal would remain weaker than the Massachusetts House bill endorsed late last month by Anthropic.

In an interview with POLITICO last week, Cesar Fernandez, Anthropic’s head of U.S. state and local government relations, called the Massachusetts House bill the strongest AI safety proposal in the nation. He drew an explicit contrast between OpenAI and Anthropic’s state lobbying efforts, noting that while OpenAI is working to bring state AI safety rules in line with those already passed elsewhere, Anthropic is pressing state legislators to enact increasingly tough safety regulations.