Massachusetts Artificial Intelligence Regulation Network

News

  • Tracking Data Center Moratoriums

    On Tuesday, Gov. Maura Healy signed an executive order requiring that any data center projects in the commonwealth obtain a community benefits agreement with the host municipality before state agencies can issue any permits or authorizations. It comes as a number of municipalities consider or have already authorized temporary or permanent moratoriums on data center construction. The order, which also directs the Department of Environmental Protection to set additional requirements for data center projects with peak electricity demand exceeding twenty-five megawatts, is a noticeable shift for an administration which has pushed for the expansion of artificial intelligence throughout the government and follows the governor’s decision in June to pause a tax incentive program for data centers.

    In recognition of the rapid growth of the data center moratorium movement and the relationship between generative artificial intelligence and data center growth, MAIRN is now tracking local data center moratoriums in Massachusetts.

  • An Act Relative to Economic Development in the Commonwealth

    On July 24th, the Massachusetts State Senate passed a major economic development bill, An Act Relative to Economic Development in the Commonwealth, after adding a number of amendments. Its text includes many policy additions and changes, including several which relate to artificial intelligence.

    The bill puts in place regulations on so-called frontier models, which are artificial intelligence models whose training required more than a threshold level of computing power. Developers of frontier models who also, along with their affiliates, have annual gross revenues greater than $500 million are required to write and maintain a frontier AI framework that applies to their frontier models and describes in detail how the developer handles catastrophic risks.1

    The bill requires the attorney general to establish a reporting mechanism for critical safety incidents2 for use by developers of frontier models and members of the public. Large frontier developers that fail to follow the reporting requirements or their own framework are subject to a civil penalty of a maximum of $1 million for a first violation and a maximum of $3 million for subsequent violations. It also provides whistleblower protections for employees of these developers.

    The bill also creates a special commission to make recommendations on the regulation of frontier artificial intelligence models. The commission will review the risks of frontier models, how risk can be measured, practices in other states, and the feasibility of using third party auditors to assess models. The commission is required to report its findings by March 1st, 2027.

    Although any regulation is welcome, we should be clear about what this bill does and does not do. It is focused on catastrophic threats from large models created by large developers. Certainly, there should be rules in place to reduce the likelihood of such events and handle them if they occur. This approach is, however, a narrow one. It does not set any rules regarding how personal data is used in training models. It does not put in place any environmental protections to manage the impact of the data centers used by those models. It does not address the effects of exposure to generative artificial intelligence on children nor the consequences of shifting decision-making authority from humans to tools driven by large language models. There are a multitude of issues that don’t necessarily arise from a single catastrophic event but are still hazards that need to be addressed.

    The aforementioned gaps in regulation are all the more striking when considering what else the bill does. Embedded in its provisions are multiple streams of funding for artificial intelligence projects. The two most significant are a $75 million program for the development and application of artificial intelligence technologies in various sectors of the state’s economy and a $100 million program for the defense sector which explicitly includes artificial intelligence as one of it’s targets. An additional $1.5 million is provided for the redevelopment of Worcester Memorial Auditorium into an artificial intelligence innovation center and entertainment facility. Public funding of artificial intelligence development should follow the creation of a serious regulatory framework for artificial intelligence. Until that is put in place, any state money being used in this manner is underwriting an unacceptable risk to the residents of Massachusetts.

    1. A catastrophic risk is defined as “a foreseeable and material risk that a frontier developer’s development, storage, use or deployment of a frontier model will materially contribute to the death of, or serious injury to, not less than 50 people or not less than $1,000,000,000 in damage to, or loss of, property arising from a single incident involving a frontier model that: (i) provides expert-level assistance in the creation or release of a chemical, biological, radiological or nuclear weapon; (ii) engages in conduct with no meaningful human oversight, intervention or supervision that is either a cyberattack or, if the conduct had been committed by a human, would constitute the crime of murder, assault, extortion or theft, including theft by false pretense; or (iii) evades the control of its frontier developer or user; provided, however, that “catastrophic risk” shall not include a foreseeable and material risk from: (A) information that a frontier model outputs if the information is otherwise publicly accessible in a substantially similar form from a source other than a foundation model; (B) lawful activity of the federal government; or (C) harm caused by a frontier model in combination with other software if the frontier model did not materially contribute to the harm.” ↩︎
    2. A critical safety incident is defined as “(i) unauthorized access to, modification of, inadvertent release of or exfiltration of, the model weights of a frontier model; (ii) harm resulting from the materialization of a catastrophic risk; (iii) loss of control of a frontier model that causes death or bodily injury or that demonstrates materially increased catastrophic risk; or (iv) instance where a frontier model that uses deceptive techniques against the frontier developer to subvert the controls or monitoring of its frontier developer outside of the context of an evaluation designed to elicit this behavior and in a manner that demonstrates materially increased catastrophic risk.” ↩︎

  • AI Encroachment into Education in Boston

    Yesterday evening, the Boston School Committee met and approved both the adoption of an artificial intelligence policy and the city’s first private school centered on artificial intelligence.

    The policy is fairly broad and leaves a number of open questions. It makes allusions to issues such as environmental impacts and overreliance, but it’s not clear how it actually addresses them. There are also no explicit references to potential harms to cognition and learning from artificial intelligence use in either the policy or the associated documentation. Notably, Google’s Gemini and Notebook LM are made available to most students in grades 7 and up by default and K-5/6 schools can request that these be made available to their students as well. The lack of any mention of research on cognitive harm as well as the wide-spread access to chatbots raises serious questions about what data was used in making the policy.

    The approved private school, Alpha School Boston, is run by a company named 2 Hour Learning and centered on a model of a two-hour core academic block. Along with the many general concerns with artificial intelligence in education, an investigation by the outlet 404 Media found a number of worrisome issues at already existing Alpha schools. Internal company documents that showed that lesson plans were sometimes faulty or presented illogical questions. Former employees voiced concern about the surveillance tools the academy used and the investigation found deficiencies in the storage and sharing of student data. In addition, the company has trained the AI models it uses by scraping data from a variety of other online courses without permission, potentially in violation of the terms of service of those platforms.

    The Alpha schools are part of a larger set of products which 2 Hour Learning has been attempting to role out across the country. Another one of the company’s schools, part of its Unbound Academy network, was rejected by the Pennsylvania Department of Education, which noted that “The artificial intelligence instructional model being proposed by this school is untested and fails to demonstrate how the tools, methods and providers would ensure alignment to Pennsylvania academic standards.”

    The policy document makes repeated references to the notion of responsible and safe artificial intelligence use. The reality, however, is that we simply do not have the data available at this time to make decisions about what constitutes such a use in classrooms and that the policy as it stands leaves many openings for students to be exposed to artificial intelligence in ways that are already known to pose problems. Hopefully the committee will elect to reevaluate its decisions in the future and vote for artificial intelligence regulations that help secure the education that Boston’s children deserve.

  • A Federal Effort to Preempt State AI Laws

    On Thursday, Representatives Lori Trahan of Massachusetts and Jay Obernolte of California released a draft of a concerning new artificial intelligence bill. If passed, the bill would greatly restrict the ability of states to pass their own regulations on artificial intelligence by preempting any state level regulations on the development of artificial intelligence models for a period of three years. It would, in effect, serve to stall out regulation at the state level during a crucial window where tech companies would be able to expand their reach largely unencumbered. A number of organizations previously signed onto a letter urging Representative Trahan to reject any bill which would preempt state laws.

    Representative Trahan represents Massachusetts’s 3rd congressional district. Negative feedback from her voters will be crucial both to preventing this bill from moving forward and for sending a signal to lawmakers that future efforts to preempt state regulations will carry a political cost. If you are a concerned constituent, you can sign this petition urging her to oppose any legislation that would override state AI safeguards or you can contact her local office by phone at (978) 459-0101.

  • Local Data Center Moratoriums

    Boston City Councilor Ed Flynn has introduced a hearing order to discuss a moratorium on artificial intelligence data centers in the city. The order, filed on April 29th, marks the third recent local effort to restrict the expansion of data centers in Massachusetts.

    In Lowell, which already has a data center, city councilors passed a one-year moratorium on further data center development in March. Additionally, residents involved with Honest Future for Lowell filed a lawsuit on April 27th against the Massachusetts Department of Environmental Protection the owner of the current data center, the Markley Group. The lawsuit challenges the July 2025 DEP’s approval of an air quality permit and an administrative consent order both tied to the addition of new diesel generators to the site.

    In Everett, the city council is considering a measure endorsed by the planning board to restrict data center development in the Docklands Innovation District, which is a 100-acre redevelopment site near the Mystic River. There is an associated petition which has has drawn more than 1,300 signatures.

    The growing deployment of artificial intelligence requires an ever-increasing number of data centers. These data centers in turn impose environmental burdens on the communities in which they are built and increase the demand for energy within and beyond their borders. Local movements, as well as state-wide ones such as the recent effort in Maine, are an important way of drawing attention to another set of costs imposed by the expansion of artificial intelligence and help ensure that the corporations that utilize these data centers aren’t able to externalize these costs.

  • What Existing Laws Govern Artificial Intelligence in Massachusetts?

    At present there are no laws in Massachusetts that explicitly regulate the use of artificial intelligence. There are some regulations in place in the form of guidance issued by government agencies.

    In 2024, the Massachusetts Attorney General’s Office issued an advisory statement on how it interpreted existing law to apply to artificial intelligence. A large part of the statement covers advertising of AI systems: suppliers can’t misrepresent the functionality, safety, reliability or performance of such systems. Additionally, the statement notes that AI technology is covered in the case of using faked audio or video to deceive a person in a business transaction. AI systems are also required to comply with preexisting rules for safeguarding personal information used by those systems. Finally, developers, suppliers and users of AI systems may not use AI systems that discriminate against residents on the basis of a legally protected characteristic.

    The Massachusetts Division of Insurance also put out a bulletin in 2024, which primarily served as a reminder that any AI systems used for making insurance decisions must comply with preexisting law. (For example not setting rates which are excessive or unfairly discriminatory). It did take the additional step of mandating that each insurance company in Massachusetts develop a written program, referred to as an AIS Program, to mitigate any adverse consumer outcomes.

    The efforts made by the Attorney General’s Office and the Division of Insurance are certainly a welcome start. There are, however, many gaps here. One of the biggest is that absent mandates for companies to engage in audits of their own systems and make that information publicly available, it can be very hard to identify if a system is in compliance with some of these regulations, for instance those protecting against discrimination. Additionally, guidance issued by agencies can shift with changing administrations in the absence of explicit language in law.

  • New Website

    The Massachusetts Artificial Intelligence Network has a new website at mairn.org

Powered by