The Trump administration has been weighing an artificial intelligence version of the Financial Industry Regulatory Authority, the private regulator that polices Wall Street’s brokers. It’s the right instinct aimed at the wrong target.
The proposal, which Treasury Secretary Scott Bessent helped develop, would create a body reporting to the Securities and Exchange Commission to vet the top AI models, Bloomberg reported in July. Google DeepMind CEO Demis Hassabis has pitched a similar plan. It covers the labs building the biggest systems. It says nothing about the AI already in consumers’ hands, and on their faces.
Stay informed.Stay ahead.
Join Washington Examiner for unlimited access to the news, analysis, and commentary that matter most.
Already a member? Log in
That’s where the next trust fight is. Meta explored facial recognition for its smart glasses. In April, 77 organizations led by the American Civil Liberties Union urged the company to drop it. In June, after reporters found unreleased code in the glasses’ companion app, Meta pulled it. Last week, it announced it is bringing Private Processing — its encrypted computing system — to the glasses.
Meta made the right call. But nobody should want an industry this important governed by sign-on letters and app teardowns. Companies can’t plan around lines they can’t see. Consumers can’t trust lines that keep moving.
Leaving it to the states produces a stack of rulebooks a startup in Austin can’t afford to navigate, and a lab in Shenzhen will never have to. The White House warned in March that a patchwork of state laws “would undermine American innovation.” But pausing state laws, as the Great American Artificial Intelligence Act would do for three years, doesn’t write the rules that replace them. That leaves a vacuum, and vacuums get filled by pressure campaigns.
The FINRA model can fill it, if Congress points it at consumer AI too. FINRA is a private nonprofit organization, funded by member fees rather than taxpayers and supervised by the SEC. Its rules are technology-neutral.
When generative AI arrived, FINRA didn’t wait on Congress. Its 2026 oversight report added a dedicated generative AI section for the first time, warning of bias and hallucinations, output that is “inaccurate or misleading, yet is presented as factual.” It tells firms to build a governance framework before deployment and to keep a human reviewing what the model produces.
That is the conservative answer to AI governance. Industry pays for it. People who understand the technology write the detailed rules and update them at the speed of the market. A federal regulator keeps the final word, which keeps it accountable to the elected government. Every firm, big or small, plays by the same rules. No new Cabinet agency. No 50-state maze.
It doesn’t sideline the states, either. State securities regulators still work fraud cases, and their national association, the North American Securities Administrators Association, has already flagged “AI washing,” scams that pitch trading tools as AI-powered. National rules, local law enforcement.
So widen the lens. Charter the body under an existing federal regulator and give it jurisdiction over consumer AI products, not just frontier models. Let it set standards, audit members, and discipline violators. Preempt state law where those national rules actually cover the ground, and leave state fraud enforcement alone. That’s preemption conservatives can defend: earned, not declared.
One line belongs in the statute itself: no one gets identified by their face without consent. That isn’t a burden on innovation. It’s the rule of law applied to a new tool. Walking down a public street has never meant surrendering your name to every stranger with a camera. Texas has required that consent from businesses for well over a decade. A national standard should do no less.
OPINION: CALIFORNIA JUST REGULATED AI AUDITS — AND FORGOT TO REQUIRE THE TRUTH
Sen. Ted Cruz (R-TX) warned last week that if China wins the AI race, the technology “will be focused on espionage and propaganda and control and surveillance.” He’s right, and that’s why the consent line matters. Beijing’s edge isn’t speed. It’s a state that never has to ask permission. If America answers by building the same capability with better marketing, we haven’t won. We’ve conceded the terms.
The race goes to the country whose AI people actually want to use. Clear rules, enforced evenly by people who know the business, are how you earn that. American companies are already building privacy into the hardware. Congress should give them a rulebook worth building too, before someone less friendly writes one for them.
Melik Abdul is a Washington, D.C.-based public affairs consultant, writer, and political analyst.
