The Future of Privacy Forum (FPF) announced on Tuesday that it will be withdrawing from the Multistate AI Policymaker Working Group that it has convened since last October. Criticism of the Forum has mostly focused on its willingness to import Euro-style tech regulations into American statehouses, a claim that FPF denies.
What has gone mostly unnoticed is that the Future of Privacy Forum suggested in its withdrawal notice that the working group has mitigated “a potential patchwork of conflicting and discordant AI laws that would be confusing for consumers and difficult to comply with.” Not only did the Future of Privacy Forum not mitigate an AI fairness patchwork, but it also actively facilitated its development as the neutral convener of the working group.
As recent research from the American Consumer Institute details, states have not delivered anything close to a uniform standard on AI fairness policy. Using standard text analysis techniques to compare all state proposals to Colorado, the only proposal currently signed by its state governor, we find that state proposals range from 12 percent similar under more stringent assumptions to 91 percent similar under more forgiving ones. In a legal context—where small differences in language and terminology can drive costly legal disputes—91 percent similar is still very different, and most states fall far short of that lofty level of similarity. After all, quirky punctuation placements have fueled decades of constitutional debates about gun ownership rights. We do not need that confusion in the AI debates.
None of this is surprising. Different state policymakers must answer to different constituencies that have different desires and expectations. Their incentives are, well, different. Ultimately getting these laws across the finish line in state legislatures will require political compromises that alter the words, meanings, and effects of state legislative proposals. Those proposals will then be subject to criticism by media outlets and experts. Responsible lawmakers should change their legislation to address the concerns of experts that convincingly warn of unintended consequences—but those changes will further complicate patchwork concerns.
Texas is a good example of this effect in action.
As TRAIGA sponsor, Representative Giovanni Capriglione (R-TX), recently described at a Texas Public Policy Foundation (TPPF) event, there have been “extensive local engagement process and local hearings … held to take input as he drafted a bill to respond to his Texas stakeholders’ concerns about AI.” As the proposal is tweaked to meet those demands, patchwork issues naturally rise.
Texas was also subjected to extensive criticisms by experts in the media extolling the wide sweeping dangers within their initial proposal. The legislature rightly amended its AI fairness bill—improving the policy effect of the law somewhat—but also further complicating patchwork concerns.
The reality is that states rely on nebulous definitions of amorphous terms that vary significantly across states. Texas, Oklahoma, and Hawaii propose entirely new state agencies to oversee compliance when others do not. The definition of a consequential decision varies drastically from state-to-state. Criminal justice, utility, and transportation decisions sometimes trigger expensive compliance burdens and other times they do not. States do not even agree on how to define key terms like artificial intelligence.
As the FPF exits the forum that it has convened, it leaves a complicated legacy behind. It may have never intended to oversee a sprawling patchwork of state AI fairness proposals, but it has facilitated that development nevertheless. Of course, much of that may have been outside of their control. After all, it has seemed that the United States was on a collision course with a state patchwork ever since Colorado mistakenly signed their own proposal into law in May 2024. But then again FPF scholars did not exactly oppose Colorado’s efforts either.
To claim that the working group has somehow mitigated patchwork concerns is also to downplay the severity of the patchwork issue still at hand. Congress could—and must—create a framework of its own to pre-empt state proposals. But that is unlikely—as in never going to happen. So, as it stands now, the most likely path to mitigate an expensive and complicated AI fairness patchwork is for the members of the Multistate AI Policymaker Working Group to swear off their collaborative efforts entirely and pass no new AI fairness laws at all. But that is also unlikely.
The departure of the Future of Privacy Forum may change the trajectory of state AI policy dramatically, but at this point it may be too little, too late. The patchwork already has too much momentum.
Logan Kolas is the Director of Technology Policy at the American Consumer Institute, a nonprofit education and research organization. For more information about the Institute, follow us on X @ConsumerPal.