Regulators are panicking. That much is clear from the flood of proposed rules, executive orders, and legislative frameworks aimed at artificial intelligence. The impulse is understandable. New technology breeds uncertainty. Uncertainty breeds fear. Fear breeds calls for action.
But here's the unpopular take: restraint, not speed, may be the smarter regulatory strategy.
The current climate reminds me of the scramble to regulate emerging industries in the past. Social media. Cryptocurrency. Ridesharing. In nearly every case, agencies moved fast, locked in rules that made intuitive sense at the time, and then spent years unwinding the consequences. Sometimes those rules helped. Often, they calcified around misconceptions and stunted beneficial innovation while missing the actual risks that materialized.
AI regulation is happening at warp speed. Multiple federal agencies are staking territorial claims. States are drafting their own rules. Congress is holding hearings. The European Union already passed its AI Act. Everyone wants to be first, which almost always means being hasty.
The problem isn't the desire to regulate. It's the timeline. We don't yet have a clear understanding of how these systems will actually behave at scale in the real world. We don't have reliable metrics for measuring fairness, transparency, or risk in ways that regulators can enforce. We don't understand which harms are genuine and which are theoretical. And we certainly don't understand which regulatory levers will actually prevent problems versus simply shifting them elsewhere.
Consider the transparency demands appearing in nearly every proposed framework. On the surface, they sound sensible. If an AI system makes consequential decisions about credit, hiring, or housing, shouldn't we understand how it works? Absolutely. But mandating disclosure of training data or model architecture before we've figured out what disclosure actually accomplishes is backwards. You end up with compliance theater: companies check boxes by releasing technical documents that neither regulators nor the public can meaningfully interpret.
The Baltimore kratom ordinance and various state-level product bans offer a useful cautionary tale here. Regulators moved quickly based on available information, locked in rules, and then faced years of legal challenges and technical disputes about what the rules actually meant. Speed felt necessary at the time. In retrospect, it created more friction than clarity.
There's also the competitive angle nobody wants to discuss openly. The U.S. is in a genuine race with China and Europe over AI development. Premature or poorly designed regulation doesn't just stifle American companies; it potentially hands advantages to foreign competitors with fewer scruples or a different regulatory philosophy. That's not an argument for the Wild West. It's an argument for getting it right rather than getting it fast.
What would restraint look like in practice? It would mean establishing mandatory reporting requirements that give regulators real data about how these systems perform. It would mean letting some harms occur, studying them carefully, and building rules around evidence rather than speculation. It would mean resisting the urge to copy Europe's approach just because it exists. It would mean Congress taking eighteen months to pass thoughtful legislation rather than six months to pass something that sounds tough.
The hardest part is that restraint looks like inaction to the outside world. Regulators don't get credit for not regulating. Politicians don't get soundbites from measured deliberation. But the cost of getting this wrong is real, and getting it right requires understanding what we're actually regulating.
The technology will evolve. The risks will become clearer. The regulatory options will improve. Waiting isn't always weakness. Sometimes it's wisdom.