by Jared Valanzola
Imagine a world in which we let an airplane manufacturer with a string of serious safety incidents write the legislation that regulates safety, not just for itself but the airplane industry at large. That world might seem far-fetched, but it’s exactly what is happening in Massachusetts's AI industry, particularly in the absence of a national regulatory framework.
If you’ve been following Beacon Hill politics during this session, you will have seen a highly concerning trend emerge. While corporate influence over our state lawmakers is nothing new, Anthropic, a company that might brand itself “responsible” but is behind models that have hacked into other organizations during testing and have proven adept at exploiting dormant bugs to take control of critical systems, has taken an active role in trying to shape the safety rules lawmakers are looking to set for their industry.
A few months ago, Anthropic publicly endorsed a plan championed by Sen. Barry Finegold to institute some of the strictest auditing standards for AI companies, requiring them to hire independent evaluators to assess potential catastrophic risks posed by larger “frontier models.” Through an amendment presented by Sen. Mike Rush, that language eventually made its way into the Senate’s version of the economic development package and saw support from several of Finegold and Rush’s peers.
So, what’s the catch? The support ginned up for Anthropic’s preferred AI safety plan is not entirely altruistic. In fact, much of the support it is seeing across both chambers is likely attributable in part to Anthropic giving more than $120,000 to the campaigns of not just those who have championed it, but also several other sitting lawmakers, the Democratic leadership’s political action committee, and the state Democratic Party as a whole.
A move like this should come as no surprise. Companies have always tried to influence the legislative process to secure favorable legislation. But it’s rare that such a blatant example of private interests muddying the legislative process actually plays out in front of our eyes.
Whether this is an attempt to maintain their “responsible” and “safe” brand or a more sinister effort to put pressure on their competitors, lawmakers must think carefully about whom they are listening to. After all, the rules we set during this legislative session for AI companies could mean the difference between a growing, vibrant AI economy and that opportunity slipping across state lines into a more favorable regulatory environment.
What we are seeing play out on Beacon Hill is just another example of why creating a federal safety standard for AI is so important. There’s a reason the vast majority of industry safety standards are set by Congress. If we let this play out Anthropic’s way, Massachusetts will both join a growing patchwork of non-interoperable laws across states and do it with language written by one of the very companies it seeks to regulate.
Thankfully, the outlook is not all grim. Representative Lori Trahan has made significant progress in this Congress by working across the aisle to find a workable path forward on AI safety policy. With both our state and national AI competitiveness determined by the decisions we make in the coming months and years, we can only hope that Trahan’s peers will join her in creating a uniform national standard that rejects the influence of companies trying to write their own rules.
Jared Valanzola is a Plymouth County Commissioner and host of The JV Team on WATD.