As Congress weighs federally pre-emptive AI regulation, states are flooding the zone with a tidal wave of short-sighted, legally dubious artificial intelligence (AI) laws. Less than four months into 2026, more than 1,700 AI-related bills are already pending nationwide, with new laws passing each month.
The volume undersells the danger. Some states are no longer considering standalone bills. Instead, they are cobbling together fragments from hundreds of unrelated AI bills into sprawling “AI mega-bills.” These bills present as one law but function as many.
This audacious approach to AI regulation reflects regulatory hubris from state policymakers who have thus far proven themselves ineffective AI regulators. Rather than attack one issue at a time, they are trying to do everything at once, over-reaching and causing a lot of damage along the way.
New York, California, Illinois, and Colorado are responsible for almost 24 percent of all AI legislative activity, but they are not alone. Many states are passing far-reaching rules on a wide range of themes: AI-enabled pricing, frontier model AI lab transparency, chatbot use, jobs and automation, data center construction, and “algorithmic discrimination.”
Until recently, most of these concerns were addressed separately and in a way states could at least try to vet and consider the unintended consequences of the legislation they were rushing across the finish line. The rise of omnibus measures in states like Florida, Connecticut, Utah, Nebraska, Tennessee, and Illinois throws caution to the wind, which only adds to the confusion associated with the growing patchwork of costly state-by-state AI mandates.
Surprisingly, it is Florida and Republican Governor Ron DeSantis that blazed this trail with the highly controversial “AI Bill of Rights,” an omnibus package of red tape filled with counterproductive rules of conduct and speech suppression. Its original definitions were broad and vague; its data regulations would have slowed innovation and invited frivolous lawsuits; and its many other rules and stipulations were often duplicative and unnecessary.
Fortunately, the Florida AI of Bill of Rights was rejected by the state general assembly, but not before the mega-bill idea had spread to Connecticut, which recently passed its own version of an omnibus AI law through the state legislature. Governor Ned Lamont intends to sign the law. Should he carefully consider the deliberations and rushed nature of the state legislature, he may arrive at a different conclusion.
The legislative process was an absolute mess. Faced with limited time and an all-encompassing bill, state policymakers pressed to get something across the finish line before the legislature gaveled out of its “short session” on May 6th. But in its haste, the legislature sacrificed prudent, deliberate policymaking.
Connecticut Senator Rob Sampson (R-Wolcott) detailed several “strike all” amendments–a process of deleting and replacing entire texts–within 24 hours, leaving some policymakers to conclude that they were unable to “understand the bill they were voting on.” With the legislative process rushed and the bill itself not fully understood, it is unsurprising that State Senator Tony Hwang, (R-Fairfield) has warned that Connecticut “may be creating a roadblock that hampers business success, business innovation.”
Other policymakers, including Minority Leader Stephen Harding (R-Harding), stressed that the proposal deals with fast-moving technology and “should be done at the federal level.” This concern is valid because many of the issues that Connecticut and other states are considering in these AI mega-bills would be better handled through a national policy framework with a more coherent and cost-effective legal vision.
There are other issues in play with these laws, some of which were highlighted in a recent lawsuit that xAI filed against the state of Colorado for its controversial AI law that it passed two years ago. That law looked to preemptively prevent “algorithmic discrimination” through confusing, open-ended mandates. Even Democratic Gov. Jared Polis argued it would, “create a complex compliance regime for all developers and deployers of AI,” through, “significant, affirmative reporting requirements.” Colorado tech innovators also sounded off against the law before it passed and warned of how it might exacerbate the state’s deteriorating business climate.
Colorado’s AI law now serves as a cautionary tale for other states not just because of its negative impact on innovation and competition, but also because it raised speech considerations and other constitutional concerns, as xAI’s new lawsuit alleges. The U.S. Department of Justice (DOJ) joined xAI’s effort in Colorado with a lawsuit of its own. “Laws like Colorado’s that force AI models to produce false results or promote ideological bias threaten national and economic security and must be stopped,” the DOJ argues. Within hours of the DOJ action, the state agreed to delay enforcement of the law until the legislature could repeal it or come up with an alternative. State lawmakers struggled to find a workable balance and finally settled on a watered down bill in mid-May.
The rise of AI mega-bills raises a new challenge: many different layers of speech suppression and legally dubious policy proposals packaged together into one sweeping law. As Colorado’s experience makes clear, addressing even one problem at a time in AI can backfire spectacularly. Addressing all of them at once is the legislative equivalent to moving fast and breaking things. It’s careless and irresponsible.
Unfortunately, some lessons are only learned the hard way. Fortunately, it’s not too late to reverse course and learn Colorado’s lessons before repeating them. But that will require state lawmakers to exercise a degree of humility and restraint absent in AI policymaking.
Omnibus AI legislation will do more harm than good. Policymakers should avoid it and instead identify how the many laws and programs that already exist in their states can address their concerns. Artificial intelligence is a general purpose technology, which means it is already regulated by existing consumer protections, civil rights laws, and other generally-applicable laws. There is no need to re-invent the wheel with technocratic mega-bills that create more problems than they solve.
Adam Thierer is a Senior Fellow for Technology and Innovation at the R Street Institute. Logan Kolas is the Director of Technology Policy at the American Consumer Institute. Follow Logan on X @Logan_Kolas