OpenAI and Anthropic Split on Massachusetts AI Safety Bill Requiring 120-Day Risk Reviews

Massachusetts Pushes Nation's Strictest State-Level AI Safety Laws

Massachusetts lawmakers are weighing what could become the toughest state-level AI regulation in the United States. The Massachusetts Senate passed an economic development bill last month that includes provisions requiring AI labs to undergo independent reviews of catastrophic risks every 120 days1

1

2

2

. These third-party evaluations of AI models would focus on potential dangers including incidents that could kill or injure at least 50 people or cause more than $1 billion in property damage1

1

. The results would be publicly released, though the state would not have authority to halt AI development based on these findings1

1

.

OpenAI and Anthropic Take Opposing Stances

The Massachusetts AI regulation has exposed a sharp divide between two leading AI companies. Anthropic has thrown its support behind the legislation, with Cesar Fernandez, the company's head of U.S. State and Local Government Relations, stating: "We applaud the Massachusetts Senate for passing the clearest and strongest AI safety legislation in the country"1

1

. Fernandez argued that more intensive third-party evaluations are necessary because "we ultimately don't think the industry should grade its own homework"2

2

. OpenAI has taken a different position, with Donnie Fowler, head of U.S. state policy and partnerships, warning that "different rules don't make things safer. They just create confusion"1

1

. The ChatGPT maker is pushing states to follow a consistent federal framework modeled after Illinois law and has cautioned that the reviews could delay rollout of cybersecurity-focused AI models designed to address the very threats lawmakers worry about1

1

.

How Massachusetts Proposal Differs From Other States

The Massachusetts bill goes significantly further than AI safety measures adopted by California, New York, and Illinois. While Illinois' artificial intelligence law requires third-party audits once a year to ensure companies follow their own safety guidelines, the Massachusetts proposal mandates independent reviews of catastrophic risks every 120 days based on external organizations' criteria rather than company standards1

1

2

2

. The legislation also requires an annual compliance review in addition to the quarterly risk assessments1

1

. Both companies have ramped up lobbying efforts, with Anthropic retaining Boston-based Tremont Strategies Group and OpenAI hiring Benchmark Strategies1

1

.

What Happens Next and Broader Implications

A six-member negotiating committee of House and Senate members is working to resolve remaining differences in the economic development bill, including the AI provisions1

1

. The legislation is expected to pass before the November election because it includes funding for projects in lawmakers' home districts1

1

. Massachusetts State Senator Barry Finegold has argued that independent testing is essential to ensure frontier AI models don't pose serious risks to the public, and if the state moves forward, other states are likely to adopt similar laws1

1

. Anthropic has demonstrated its commitment to AI safety through transparency and oversight, recently donating another $20 million to pro-regulation group Public First Action, bringing total contributions to $40 million2

2

. States introduced 260 AI-related bills in the first six months of 2025 and passed 22 into laws, signaling growing momentum for state-level AI regulation even as companies debate whether a federal framework would better serve AI safety goals2

2

.