On February 15, 2025, Equity & AI published "Biden's AI Protections Revoked: Trump's New Executive Order and the Future of AI Equity." This is the follow-up.
On January 23, 2025, President Trump signed Executive Order 14179 and revoked his predecessor's landmark AI governance directive. The Biden order had required safety testing for frontier models, mandated bias audits in federal hiring systems, directed agencies to establish Chief AI Officers, and named housing, credit, and employment as high-stakes domains requiring scrutiny. It was the most comprehensive federal AI policy ever issued. It was gone by noon on inauguration day.
The replacement order was titled "Removing Barriers to American Leadership in Artificial Intelligence." Its promise was embedded in that framing: the barriers were coming down, American leadership was coming up, and something better was on its way. When a government dismantles a framework, the implicit contract with the public is that it intends to build something in its place. Fourteen months later, we can evaluate that contract.
Here is what we got.
Eleven Months of Nothing
From January 23, 2025 through December 10, 2025, there was no federal AI framework. No governance standard. No accountability floor. No replacement for what had been revoked. During those eleven months, algorithmic systems continued making consequential decisions about employment, credit, housing, and healthcare for millions of people. State legislatures scrambled to fill the vacuum with their own laws. And the federal government, which is the only institution with the authority to set a national standard, was absent.
What arrived during this period was not governance. In July 2025, the administration issued an executive order titled "Preventing Woke AI in the Federal Government," restricting equity-focused AI practices within federal agencies. If the January revocation had raised questions about the administration's priorities, the July order answered them.
What Arrived in December
On December 11, 2025, President Trump signed Executive Order 14365, "Ensuring a National Policy Framework for Artificial Intelligence." The title sounds like governance. The contents are something different.
The order does not establish consumer protections for people affected by AI decisions. It does not create transparency requirements. It does not set standards for the domains where algorithmic discrimination has been most documented — hiring, lending, housing, healthcare. It establishes a litigation task force.
The Department of Justice was directed to form an AI Litigation Task Force, operational by January 10, 2026, with a specific mandate: challenge state AI laws in federal court. The grounds for challenge include unconstitutional burdens on interstate commerce, federal preemption, and any other basis the Attorney General deems appropriate. The order also conditions $42 billion in broadband infrastructure funding, allocated under the Broadband Equity Access and Deployment program, on states repealing AI regulations the administration considers onerous.
The federal government spent eleven months producing nothing, and then produced a mechanism to prevent states from producing anything either.
The Deadlines That Passed
Executive Order 14365 set its own internal deadlines. The Secretary of Commerce was required to publish a comprehensive evaluation of existing state AI laws by March 11, 2026. The Federal Trade Commission was required to issue a policy statement on AI deception standards by the same date. The Department of Commerce was required to issue a public notice on BEAD funding conditions by March 11, 2026.
March 11 passed. As of this writing, the Commerce evaluation has not been published. The FTC policy statement has not been issued. The funding notice has not been issued. An administration that declared state AI regulations a national emergency, that created a litigation task force to challenge them, and that issued specific deadlines for its own agencies to act, did not meet those deadlines.
This is the record of an administration that does not intend to govern AI. It intends to ensure that no one else does either.
The March Blueprint
On March 20, 2026, the White House released what it called a National Policy Framework for Artificial Intelligence — a four-page legislative blueprint directing Congress to establish a uniform federal approach. The framework identifies six objectives: protecting children online, safeguarding against AI harms, respecting intellectual property, preventing AI censorship, promoting innovation, and developing an AI-ready workforce.
The document contains no binding requirements. It creates no enforceable standards. It leaves what analysts have identified as wide regulatory gaps in bias standards, adult data privacy, and transparency mandates. It instructs Congress to act this year. That is not a policy framework. It is a referral.
Fourteen months after revoking the most comprehensive federal AI governance order ever issued, the administration's answer is four pages and a request for Congress to figure it out.
The States Trying to Fill the Gap
This is where Colorado matters, not as the central story but as evidence of what the pattern looks like in practice. When we published our examination of Colorado Senate Bill 189 earlier this month, we documented how industry lobbying stripped the explanation requirement from the first-in-the-nation AI accountability law. What that piece could not fully address was the federal context in which that lobbying operated.
David Sacks, Trump's Special Advisor for AI and Crypto, called Colorado's law "probably the most excessive" in December 2025. The Colorado AI Act, which requires companies to explain how their AI makes consequential decisions about people's lives, is now a target of the federal AI Litigation Task Force. The state delayed its own law from February 2026 to June 2026 under this pressure. Thirty-six state attorneys general opposed the federal moratorium on state AI laws, warning that it would freeze states' ability to respond to emerging risks.
This is what the vacuum looks like when it is weaponized. States that tried to protect their residents are being sued. Federal funding is being conditioned on their compliance. The accountability laws that communities fought for are being described by federal officials as obstacles to innovation.
There Is Still No Framework
As of May 2026, there is no federal AI governance framework. There is no national standard for how AI systems must treat the people they affect. There is no federal requirement that a company using AI to deny your loan application, screen your job candidacy, or determine your housing eligibility must be able to explain how it made that decision.
What exists is a deregulation order, a litigation task force, a funding threat, a set of missed agency deadlines, and a four-page referral to Congress.
In February 2025, this publication documented what was lost when the Biden order was revoked. We noted then that accountability for the people most likely to be harmed by consequential AI systems had been traded for the framing of national competitiveness. Fourteen months later, the trade is clearer. What was promised was a national standard. What was delivered is a federal apparatus designed to ensure that no standard, state or federal, can hold AI systems accountable to the people they affect.
The promise was governance. What arrived was its opposite. And there is still nothing in its place.