The AI Regulation Ban is Out of the Big Beautiful Bill and What's Next

It was big. It was beautiful. It was…4 in the morning. During another marathon session to get the One Big Beautiful Bill Act (OBBB) through the Senate at the behest of President Trump, a major vote was taken. The vote wasn’t close. Bleary eyed senators were not voting on the bill itself, but whether to remove a proposed ten-year ban on artificial intelligence (AI) regulation by states, the brainchild of Texas senator Ted Cruz. The result? 99-1. Against. The AI regulation ban failed spectacularly and now many are asking what it means. Pure capitalists may say the removal of the language opens the door for industry and innovation stifling innovation. States’ rights advocates no doubt cheered. As with everything in laws and policies, nothing is that simple or clear. The removal of the AI regulation language will have nuanced meaning for industry, state governments, developers, and consumers. Broadly, this move puts the AI regulation conversation on par with the privacy conversation, which is to say that it is being deferred to states. Another way to understand this move is that there are opportunities to create new markets. Without a monolithic federal AI bill (like the EU AI Act ), we will not have a monolithic AI market. There will be challenges in a more patchwork legal and regulatory environment (like we currently have in privacy) but there are opportunities as well. The removal of the state AI regulation moratorium language is almost certainly being cheered by anti-regulation advocates, but for the wrong reasons. Subscribe now Appetite for AI Regulation When I worked at DHS, I used to go to painstaking lengths to not use the “R” word when I engaged with industry. As a technology policy maker at DHS, I had ZERO authority to impose regulations on literally anything, but so sensitive was the mere thought, that I had to choose my words very carefully. This is because regulation is a dirty word to most Americans and is looked upon as a stifler of innovation. Senator Cruz undoubtedly had this in mind when he drafted the AI regulation ban language for the OBBB. At some point in the night, probably when he was talking to Senator Marsha Blackburn, he realized it wasn’t going to happen. The ensuing vote left no doubt. 99-1. Hard to argue with that. In this case, a 99-1 vote against an AI regulation ban means that there is at least some level of interest in AI regulation on a broad enough scale to be significant. The Senate could have chosen to pass this piece of the bill, and it frankly would have probably been easier than what they did. But the resounding 99-1 vote against the measure is telling. This is not to say that large amounts of stifling and cumbersome regulation is inbound on AI, almost certainly not. However, it is to say that there is a recognition that a full ban isn’t the answer either. The EU took preemptive action when it passed the EU AI Act. The US tends not to roll like that. However, this signal should be observed by the market. Instead of fighting for ZERO regulation on AI, the perspective is different. Now, the perspective is around what AI regulation and how it will be applied. We’ve seen a primer on how this may play out with privacy legislation. The bottom line is that the senators voting on the bill could have chosen to leave the ban in and didn’t. Even House members later talked about the AI regulation ban as an issue that, had they fully understood it, would have potentially swayed their vote. That’s about as bipartisan a signal as we can hope for. The door is open for AI regulation at the state level. Share Frontier Foundry Substack Monolith or Patchwork? Most prominent of the signals sent to the AI industry and its consumers by senators pulling an all nighter as if they were cramming for a history exam is that we can all forget about big, sprawling federal legal actions on AI for at least the foreseeable future. The proposed ten-year ban on AI regulation by states was likely an overplay of a hand that is going to have interesting knock-on effects. For years, privacy advocates have bemoaned the lack of federal privacy legislation. Basically, a US version of GDPR . For years, no such law was on the horizon. Then came the states. Today, 19 states have data privacy laws. That’s the definition of a patchwork. More states are considering privacy legislation, which will make the patchwork more…patchy. Image Credit AI laws and regulations are likely to follow the same path in the wake of the Senate’s all-nighter, so the decision is effectively made. The federal government is not going to step in on AI regulation in a meaningful way, even if it is to prevent it at a state level. Leave a comment Regulation as an Enabler Cover your ears because I’m about to use the “R” word. Think about industries that we regulate today. Here are some examples: Air travel Financial markets Nuclear technology Why do we regulate industries? We regulate industries that we collectively agree we cannot allow to fail. What is the acceptable number of commercial airline crashes per year? Easy. Zero. What about the acceptable number of bank failures or the acceptable number of nuclear accidents? Zero. In establishing that we will tolerate zero failures in these areas, we also created a demand for professionals and technology to enable those regulations. The regulations do impose additional steps on those in the regulated markets but they also open new markets for talent and technology to enforce and audit those regulations. Just with most things, the truth is in the middle. Is even the mention of the “R” word enough to kill industries and doom us all to Hoovervilles? Image Credit No. Is a lack of regulation going to get us all our own private jet? Also no. When it comes to AI regulation, we need to ask ourselves if AI is a technology of which we can tolerate failures. What is the acceptable number of AI failures per year? Right now, that number is high. It would be easier to count it by day. The point is that there are economic opportunities even in regulation. One law whose life was spared at 4am was New York’s RAISE Act . The act will establish safety and transparency standards for powerful AI systems that cost over $100 million to train. It requires developers to implement safety protocols, report incidents, and undergo third-party audits to prevent potential harms from advanced AI technologies. This is the kind of law that would have been invalidated by the AI regulation ban had it stayed in place. Wherever you personally come down on regulation as a concept and specific regulation of AI is your business. What’s clear is that the Senate could have banned AI regulation for a decade and chose not to. What’s also clear is that laws like New York’s RAISE Act are seeking to create safety around certain types of AI and that those laws will also open economic opportunities. This is not to say that all regulation is good…or bad. It is to say that we need to read the right signals from the OBBB passage earlier in the week. We can’t see this as partisan infighting or as a comment on President Trump’s agenda. Instead, it is a signal of where AI regulation is going and what it will mean for industry. I’ve said the “R” word too much for even my taste. Old habits die hard. Nick Reese is the cofounder and COO of Frontier Foundry and an adjunct professor of emerging technology at NYU. He is a veteran and a former US government policymaker on cyber and technology issues. Visit his LinkedIn here .