OpenAI and Google Push Back on Massachusetts AI Safety Rules Championed by Anthropic
Key Takeaways
- •The Massachusetts Senate passed a bill in late July 2026 requiring frontier AI developers—those with more than $500M in AI-derived revenue or over $1B in R&D spending—to publish safety frameworks focused on catastrophic risks.
- •Anthropic is advocating for independent risk assessments every four to six months, a stricter schedule than any other current state-level AI regime, while OpenAI and Google oppose the tougher provisions.
- •OpenAI prefers a narrower approach modeled on Illinois' law, favoring annual third-party audits over semiannual independent risk evaluations, citing concerns that fragmented state rules could delay beneficial AI releases.
- •Anthropic made campaign contributions exceeding $120K to Massachusetts Democrats in August 2026, while the Greater Boston Chamber of Commerce warned restrictive rules could deter AI investment in the state.
- •As of late August 2026 the bill remains in bicameral negotiations, and with no comprehensive federal AI safety law in place, its outcome could serve as a template for other state legislatures.

The largest players in artificial intelligence remain divided over how their own industry should be regulated. Massachusetts has become the newest front in the battle over state-level AI safety rules, with OpenAI and Google opposing provisions that Anthropic is actively promoting as the strongest AI safety proposal at the state level anywhere in the country.
In late July 2026, the Massachusetts Senate passed an economic development bill containing AI guardrails aimed at frontier developers — specifically companies that generate more than $500M in AI-derived revenue or spend over $1B on research and development. The legislation requires those companies to publish safety frameworks focused on catastrophic risks, grants enforcement authority to the state attorney general, and launches a study into independent evaluations of AI systems. The thresholds mean the rules would apply to only a small number of the largest AI developers rather than the broader AI industry.
Three labs, two camps
Anthropic wants the rules to go further. The company is advocating for independent risk assessments every four to six months, a schedule that would make Massachusetts' requirements substantially stricter than any other state-level regime currently in force. Anthropic has described its preferred version as the most robust AI safety legislation in the nation. The split is notable because Anthropic, OpenAI, and Google are the leading developers of frontier AI models, and their disagreement illustrates that even the companies building the most advanced systems do not share a common position on how tightly they should be overseen.
OpenAI takes a different view. It supports a narrower version of the bill and has warned that aggressive state-by-state regulation risks producing a patchwork of inconsistent rules. That fragmentation, OpenAI argues, could delay the release of beneficial AI models, including cybersecurity tools. The company instead favors an approach closer to Illinois': annual third-party audits rather than independent risk evaluations twice a year.
Google has aligned with OpenAI in opposing the stricter provisions.
Money talks in Massachusetts
Anthropic's advocacy extends beyond policy papers. In August 2026, the company made campaign contributions exceeding $120K to Massachusetts Democrats.
On the other side, the Greater Boston Chamber of Commerce has warned that overly restrictive regulations could deter AI investment in the state.
Alongside the safety rules, the bill earmarks $75M for AI development. As of late August 2026, the legislation remains in bicameral negotiations, meaning the final version could differ substantially from what the Senate approved. Whatever emerges from conference committee will set a benchmark other states are likely to study as they weigh their own bills.
The bigger regulatory chess match
Since California's high-profile AI safety bill was vetoed by Governor Newsom in 2024, state legislatures across the country have been drafting their own approaches. With no comprehensive federal AI safety law in force, states have become the primary testing ground for binding AI oversight in the United States, and the outcome in Massachusetts could shape the template other legislatures adopt.
The enforcement mechanism also carries weight. Empowering the state attorney general to act on violations of published safety frameworks introduces legal risk beyond voluntary commitments or industry self-regulation. Companies operating in Massachusetts would have to treat their published safety frameworks as binding documents.