Edition 007: Who Writes the Rules?

When we talk about how the federal government governs anything, we are usually talking about laws — legislative bills introduced by Congress, debated, amended, and then signed the President; rules built to outlast the administrations that wrote them. That is how American government has handled the critical issues, like civil rights, securities regulation, food and drug safety, and environmental protection. While none of these is perfect, each rests on a law that Congress passed, the President signed, and no one alone can erase.

Artificial intelligence has no such law. Congress has not enacted any comprehensive AI legislation.  In fact, it has not even enacted any narrow AI legislation. Here, I should be clear that the federal government has not been entirely silent. The FTC has brought enforcement actions for AI-driven deception.[1] And the SEC has flagged AI governance as an examination priority.[2] But there are not AI laws — they are pre-existing federal authorities being applied to AI conduct, and they are the closest thing to federal AI rules the country currently has.

***

This is more than a disappointment. It is the abdication of a constitutional function at exactly the moment that function is needed most. I say that as someone with a Texas-sized skepticism of regulation. I have spent a career arguing that the burden of justification belongs on the rule, not on the regulated. I know from the inside what federal rules cost American businesses. My conclusion here is not the reflex of someone who wants more rules. It is the conclusion of someone who has reached for that answer last, and reached it anyway. The technology that is being deployed in hiring decisions, in medical diagnoses, in courtrooms, in classroom instruction, and in the information that ordinary Americans see every day is being deployed under no federal AI rules at all.

So, this edition is about executive orders because executive orders are what is left when Congress has failed.

***

In 2023, President Biden signed Executive Order 14110, Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence.[3] It was the most comprehensive action the federal government had ever taken on AI, directing more than fifty federal agencies to carry out over a hundred specific tasks. What the order said, in essence, was that before these systems were used to screen a job applicant, evaluate a loan, or make decisions in a hospital — before they reached into the parts of life where they could change a person’s future — someone in government was supposed to set the rules for how they were tested, audited, or disclosed to the people whose lives they impacted.

The order reached into the places AI had already begun to touch quietly — hiring, housing, healthcare, banking, education, transportation, and criminal justice. Federal agencies were directed to issue guidance on algorithmic discrimination in employment, in tenant screening, and in access to credit. The Department of Health and Human Services was directed to examine AI in clinical care. The Department of Justice was directed to examine AI in the criminal legal system. The agencies that protect consumers — the Federal Trade Commission, the Consumer Financial Protection Bureau, and civil rights offices across the federal government — were encouraged to use their existing authorities against fraud, discrimination, and threats to privacy enabled by AI.

The order reached the workplace, too, and it was unusually direct about what was happening there. Workers were being surveilled without limit. Their rights were being eroded. The quality of their jobs was deteriorating. Their livelihoods were at risk of displacement. It reached the most powerful AI models, too, requiring the companies building them to notify the government when they began training an AI model above a defined size, as measured by computing power, and to share the results of their safety tests before release. It directed the National Institute of Standards and Technology to write the standards that those tests should follow. And it told every major federal agency to appoint a Chief AI Officer and to take responsibility for what their algorithms decided.

What the order did not do, and could not do, was regulate the AI industry on its own. An executive order is the President directing the executive branch. Essentially, it is the executive branch of the government talking to itself. The actual rules would still have to be written by agencies in the executive branch, through notice and comment, on the timetable that real rulemaking requires. Executive Order 14110 set that work in motion. It pointed the federal government at the problem and told it to begin to work on the solution.

That, on its own, was something the country had never seen before for AI. It lasted less than fifteen months.

***

Before the inaugural crowd had left the mall, it was gone. Signed away on January 20, 2025, in a stack of orders titled “Initial Rescissions of Harmful Executive Orders and Actions,” on a morning when the cameras were pointed elsewhere. Three days later, President Trump signed Executive Order 14179, Removing Barriers to American Leadership in Artificial Intelligence,[4] the replacement policy. The order establishes a U.S. policy of sustaining and enhancing the United States’s “global AI dominance” to promote “human flourishing, economic competitiveness, and national security.” It frames the goal as developing AI systems “free from ideological bias or engineered social agendas,” and revokes “certain existing AI policies and directives that act as barriers to American AI innovation.”

Regulatory burden is always a concern, but a serious argument should address the issues that impact people—safety, fraud, discrimination, and threats to privacy, among others. This order did none of that, but it did direct the development of an AI Action Plan.

***

In July 2025, the White House released Winning the Race: America’s AI Action Plan. [5] The plan sets forth three main pillars: accelerate AI innovation, build AI infrastructure, and lead international AI diplomacy and security. Its first policy section is titled “Remove Red Tape and Onerous Regulation,” and that theme—that the primary obstacle to American AI leadership is regulation itself—is woven through every pillar that follows.

The plan recommends reviewing “all Federal Trade Commission (FTC) investigations commenced under the previous administration to ensure that they do not advance theories of liability that unduly burden AI innovation.” It directs federal agencies to limit AI-related funding to states whose regulatory regimes may “hinder the effectiveness” of those funds. It calls for revising the NIST AI Risk Management Framework to eliminate references to “misinformation, Diversity, Equity, and Inclusion, and climate change.” Absent are requirements regarding safety testing, civil rights and anti-discrimination protections, privacy and consumer safeguards, worker displacement protections, climate considerations, equity provisions, compute reporting requirements, healthcare AI oversight, and more. The Trump administration replaced a risk-focused regulatory regime with a short deregulatory directive aimed at clearing obstacles to AI development.

The Action Plan purports to be a strategy for winning the AI race. It is not a plan for protecting the people the race runs over.

***

So the answer to the question of who writes the rules of American AI right now is whoever signed the last executive order, until someone else signs the next one. The question then becomes where in the American system is AI is being governed. The answer, for the moment, is the states — and that is what the next edition addresses.

***

[1] United States, Federal Trade Commission. “FTC Announces Crackdown on Deceptive AI Claims and Schemes.” Federal Trade Commission, 25 Sept. 2024, www.ftc.gov/news-events/news/press-releases/2024/09/ftc-announces-crackdown-deceptive-ai-claims-schemes. Accessed 6 May 2026.

[2] United States, Securities and Exchange Commission, Division of Examinations. Fiscal Year 2026 Examination Priorities. 17 Nov. 2025, www.sec.gov/files/2026-exam-priorities.pdf. Accessed 6 May 2026. Accessed 6 May 2026.

[3] United States, Executive Office of the President [Joseph R. Biden]. Executive Order 14110: Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. 30 Oct. 2023. GovInfo, www.govinfo.gov/content/pkg/DCPD-202300949/pdf/DCPD-202300949.pdf. Accessed 6 May 2026.

[4] United States, Executive Office of the President [Donald J. Trump]. Executive Order 14179: Removing Barriers to American Leadership in Artificial Intelligence. 23 Jan. 2025. The White House, www.whitehouse.gov/presidential-actions/2025/01/removing-barriers-to-american-leadership-in-artificial-intelligence/. Accessed 6 May 2026.

[5] United States, Executive Office of the President. Winning the Race: America's AI Action Plan. The White House, July 2025, www.whitehouse.gov/wp-content/uploads/2025/07/Americas-AI-Action-Plan.pdf. Accessed 6 May 2026.