Anthropic Breaks With Peers on Massachusetts AI Safety Bill

Anthropic has reportedly broken with other tech giants over a new Massachusetts AI regulation.

The proposal before that state’s senate would require artificial intelligence (AI) developers to hire independent evaluators to assess their models for the potential catastrophic risks every four months, The Information reported Thursday (Sept. 3).

According to that report, the Massachusetts proposal could establish a new standard for AI regulation, something states are pursuing amid a slowdown in federal legislation.

This is also happening at a time of increasing concern about rogue AI agents, following an attack by OpenAI agents on AI platform Hugging Face, the report added.

Anthropic and safety groups support the AI language in the legislation — part of a larger economic development bill — while companies like OpenAI and Google oppose it.

When the Massachusetts Senate passed its version of the legislation, Anthropic Head of U.S. State and Local Government Relations Cesar Fernandez wrote in a post on X that it “raises the bar on AI safety while ensuring that innovation is able to continue.”

Anthropic, which has previously supported other state-level artificial intelligence safety measures, is joined by AI safety organizations including Encode and Secure AI Project. The latter group’s policy director, Scott Wisor, told The Information that they’ve recently sent a letter of support to the members of the conference committee.

Meanwhile, OpenAI has noted its objection to the AI language in the Massachusetts bill, the report said. In letters last month to lawmakers, OpenAI policy executives wrote that Massachusetts should consider a requirement similar to an Illinois bill, which calls for annual, third-party safety audits. That letter added that “fragmented oversight” around the country would make for an “unhappy result.”

As covered here earlier this week, other state AI regulations include chatbot safeguards that Rhode Island enacted in June. Lawmakers in California are advancing measures addressing audits, chatbots and privacy, while the counterparts in Utah and South Carolina are renewing children’s online safety efforts.

These moves are happening, lawmakers increasingly see fighting AI the industry as an electoral advantage, PYMNTS wrote.

“Washington’s inability to establish a national framework has also weakened industry demands that states stand down,” that report added.

“Congress rejected a proposed 10-year moratorium on state AI laws in 2025 after opposition from governors, attorneys general and state lawmakers. A subsequent attempt to attach preemption language to defense legislation also failed.”