Federal AI policy · Executive Order 14179 · Executive Order 14365
Federal AI Governance — A Maintained Timeline
A maintained record of the federal executive actions, agency guidance, and legislative developments shaping AI governance in the United States — each entry dated, status-marked, and sourced. As of July 23, 2026.
Summary
Federal AI governance in the United States is the evolving set of executive actions, agency guidance, and legislative proposals that shape how organizations build, deploy, and oversee artificial intelligence. It is not a single statute but a sequence — orders issued and rescinded, guidance proposed, deadlines set — which is why a current position is worth more than a snapshot. This page is maintained: it records the developments that matter to boards, general counsel, and compliance leaders, each with its date and status, and is updated as the federal posture changes.
Most asked
Does the federal push mean state AI laws no longer apply?
No. As a general matter, preemption of a state law requires an act of Congress or a court ruling. The December 2025 national-framework executive order directs federal agencies to contest state AI laws, and the Justice Department has intervened against Colorado’s — but the Colorado, New York, Illinois, and California duties remained operative when those actions issued. The durable posture is documented oversight that survives whichever text is in force, not reliance on any single deadline.
What changed in federal AI policy in 2025?
Two shifts. In January 2025, federal posture moved from managing AI risk — the approach of the now-rescinded 2023 order — to accelerating AI adoption, set out in EO 14179 and the July 2025 AI Action Plan. Then, from December 2025, the federal government largely stopped writing comprehensive AI rules of its own and began contesting the states’ rules, through an executive order, a Justice Department task force, and agency proceedings at the FTC and FCC.
Is the Colorado AI Act still in effect?
No — the original act never took effect at all. Enforcement was stayed in April 2026 pending the litigation the Justice Department joined, and on May 14, 2026 Colorado repealed and reenacted the act (SB 26-189), replacing the algorithmic-discrimination duty with a narrower disclosure regime for automated decision-making technology that becomes operative January 1, 2027. It is the clearest illustration of why a compliance program should be built to demonstrate oversight rather than to a single statutory text.
Overview
Two shifts organize everything below. The first is a change of federal posture, in January 2025, from managing AI risk to accelerating AI adoption. The second, a year later, is subtler and more consequential for planning: the federal government stopped writing its own comprehensive AI rules and began contesting the states’ rules instead. The practical result for anyone building a compliance program is that the obligations themselves are now in motion — a deadline you plan around in March can be suspended by litigation in April. The durable response is documentation and oversight that survive whichever text is in force, a point the Colorado sequence at the top of this timeline makes concretely.
The entries run newest first. Each states what the development does — as a matter of sourced fact — and, separately, what it means for a board or a compliance leader, which is Techné AI’s analysis.
2026
July 1, 2026 — The FTC proposes an “AI accuracy” policy statement
Proposed · comment period open.
What it does. The Federal Trade Commission, by a 2–0 vote, sought public comment on a proposed statement that distorting AI outputs to undisclosed ends can be a deceptive practice under Section 5 of the FTC Act — and that state laws coercing alterations to AI outputs, naming Colorado’s AI Act, are “impliedly preempted” where they conflict with the federal scheme. Comments are due July 31, 2026 (docket FTC-2026-0859).
For a board. The federal preemption argument now has an agency vehicle and a public record, not just an executive order. Organizations operating under state AI-output or disclosure duties should watch whether the final statement reaches beyond Colorado.
Source: Federal Trade Commission — press release (July 1, 2026)
May 14, 2026 — Colorado repeals and reenacts its AI Act (SB 26-189)
Enacted.
What it does. Under the federal challenge, Colorado repealed and reenacted its AI Act before the June 30, 2026 effective date arrived. SB 26-189 drops the reasonable-care duty to prevent algorithmic discrimination, the impact assessments, and the risk-management program, substituting a narrower disclosure regime for automated decision-making technology — notice before use in consequential decisions, post-adverse-decision disclosure, and human-review rights — effective January 1, 2027, enforced exclusively by the Attorney General.
For a board. This is the timeline’s thesis in a single event. A duty many multi-state employers spent 2025 preparing for was contested federally and then rewritten by the state itself — before it ever became operative. Programs built to a specific statutory text are fragile; programs built to demonstrate oversight are not.
April 24, 2026 — DOJ intervenes against the Colorado AI Act; enforcement suspended
In force.
What it does. The Department of Justice intervened in xAI’s lawsuit challenging Colorado’s AI Act, and enforcement of the Act was suspended pending the litigation — the first federal action against a specific state AI statute under the national-framework policy.
For a board. The preemption push works through litigation, not a stroke-of-the-pen override. That makes it slower and contestable, and it makes a defensible internal record more valuable than any single state deadline.
Source: U.S. Department of Justice, Office of Public Affairs; Axios
March 11, 2026 — EO 14365’s 90-day deadlines fall
Pending.
What it does. The Commerce Department’s evaluation of “onerous” state AI laws, the FTC policy statement, and the FCC proceeding all came due in the same window. The FTC’s item matured into the proposed statement above; the Commerce evaluation’s publication and contents are not independently confirmed in this record.
For a board. The Commerce list, when public, functions as an early-warning map of which state obligations the federal government considers most contestable — useful for prioritizing, not for concluding that any named law has stopped applying.
Source: Butzel Long; Baker Botts
January 9, 2026 — DOJ AI Litigation Task Force established
In force.
What it does. The Department of Justice stood up the task force directed by EO 14365, charged with challenging state AI laws inconsistent with federal policy. No suits were filed at creation; action awaited Commerce’s referrals.
For a board. This is the instrument that later reached Colorado — the moment the preemption policy acquired an enforcement arm.
2025
December 19, 2025 — New York enacts the RAISE Act
Enacted.
What it does. New York adopted frontier-AI safety legislation, one of the state measures the national-framework policy was positioned to test.
For a board. A live example of the state duties that persist despite the federal push; relevant to any multi-state organization tracking which obligations still bind.
December 11, 2025 — EO 14365, “Ensuring a National Policy Framework for Artificial Intelligence”
In force.
What it does. Directs the federal government to press for one national AI framework and to challenge conflicting state laws. Its deadlines: an AI Litigation Task Force within 30 days; a Commerce evaluation of “onerous” state laws within 90 days; an FCC proceeding on a federal reporting standard within 90 days; an FTC policy statement within 90 days; and agency review of whether to condition federal funding, including BEAD, on a state’s AI posture.
For a board. The order does not, by itself, invalidate any state law — preemption generally requires Congress or a court — so Colorado, California, New York, and Illinois duties remained in force when it issued. Its force runs through the task force, the agency proceedings, and the funding lever, each of which took months to reach a specific statute.
December 8, 2025 — The state-AI moratorium is omitted from the FY2026 NDAA
Enacted.
What it does. Congress released the defense authorization without the proposed moratorium on state AI laws — the second failed attempt to preempt state AI regulation by statute. The administration announced the “ONE RULE” executive order the following day.
For a board. Broad preemption still lacks a law behind it. That is why the federal challenge proceeds by executive and litigation means, which are narrower and can be contested.
Source: StateScoop
July 23, 2025 — America’s AI Action Plan and three executive orders
In force.
What it does. The administration’s AI strategy, released with three same-day orders: Preventing Woke AI in the Federal Government (federal language-model procurement must emphasize accuracy and ideological neutrality); Accelerating Federal Permitting of Data Center Infrastructure; and Promoting the Export of the American AI Technology Stack.
For a board. The plan set the “one national framework, minimal friction” thesis that the December order later turned against state law, and the procurement-neutrality order prefigured the FTC’s 2026 “AI accuracy” theory. Directly binding on federal agencies and their vendors; a signal, not a mandate, for other employers.
Source: Akin; the White House
April 3, 2025 — OMB resets federal AI use and acquisition (M-25-21, M-25-22)
In force.
What it does. Two memoranda replaced the prior administration’s federal AI guidance — one on how agencies use AI, one on how they buy it.
For a board. These bind federal agencies, not private employers. Their relevance is as the template — chief AI officers, risk practices for high-impact systems — against which contractors and vendors are increasingly measured.
Source: Hunton Andrews Kurth
January 23, 2025 — EO 14179, “Removing Barriers to American Leadership in Artificial Intelligence”
In force.
What it does. Sets current federal AI policy — sustaining U.S. leadership, removing regulatory barriers — and directed the work that became the AI Action Plan. The prior AI order was revoked three days earlier, on January 20, 2025.
For a board. The hinge from the previous federal model to the current one. Federal guidance issued after this date reflects the acceleration posture, and reference programs built to the earlier order need re-basing.
Source: The American Presidency Project
2023
October 30, 2023 — EO 14110 (the prior administration’s AI order)
Rescinded.
What it does. The foundational federal AI executive order, directing safety, security, civil-rights, and reporting actions across agencies.
For a board. Named here because many governance programs still cite it. It was rescinded in January 2025; references to it should be updated.
Source: Federal Register — EO 14110 (88 FR 75191) (citation pending confirmation)
What this timeline is for
A federal record is only half of a compliance picture — the state duties are where most employment and consumer obligations actually live, and those are tracked separately in our Briefings. What the federal timeline gives a board is the weather system above those duties: the direction of federal pressure, the instruments in play, and the reason a fixed compliance target is now a moving one. Read it to calibrate posture, not to conclude that any specific obligation has lifted.
Related references
- Workplace AI-Use Policies — What Belongs in One — the internal-governance companion to this external record: what a durable AI-use policy contains while the law moves around it.
- The New York RAISE Act Reference — a closer read of one of the state frontier-AI duties that persists despite the federal preemption push.
- Multi-Jurisdictional AI Compliance: Building a Unified Framework — how to build one program that absorbs a moving federal-and-state map rather than parallel programs that diverge.
- BoardSight Intelligence — the maintained research library for directors and officers carrying AI oversight and D&O exposure.
Khullani M. Abdullahi holds a Juris Doctor and is the founder of Techné AI. Techné AI provides advisory services and does not provide legal advice or assurance services within the meaning of the AICPA Statements on Standards for Attestation Engagements.
Frequently asked questions
- Does the federal push mean state AI laws no longer apply?
- No. As a general matter, preemption of a state law requires an act of Congress or a court ruling. The December 2025 national-framework executive order directs federal agencies to contest state AI laws, and the Justice Department has intervened against Colorado’s — but the Colorado, New York, Illinois, and California duties remained operative when those actions issued. The durable posture is documented oversight that survives whichever text is in force, not reliance on any single deadline.
- What changed in federal AI policy in 2025?
- Two shifts. In January 2025, federal posture moved from managing AI risk — the approach of the now-rescinded 2023 order — to accelerating AI adoption, set out in EO 14179 and the July 2025 AI Action Plan. Then, from December 2025, the federal government largely stopped writing comprehensive AI rules of its own and began contesting the states’ rules, through an executive order, a Justice Department task force, and agency proceedings at the FTC and FCC.
- Is the Colorado AI Act still in effect?
- No — the original act never took effect at all. Enforcement was stayed in April 2026 pending the litigation the Justice Department joined, and on May 14, 2026 Colorado repealed and reenacted the act (SB 26-189), replacing the algorithmic-discrimination duty with a narrower disclosure regime for automated decision-making technology that becomes operative January 1, 2027. It is the clearest illustration of why a compliance program should be built to demonstrate oversight rather than to a single statutory text.
- Does EO 14365 preempt state law on its own?
- No. The order directs the federal government to seek one national framework and to challenge conflicting state laws, but it does not by itself invalidate any statute. Its force runs through the DOJ litigation task force, the FTC and FCC proceedings, and the prospect of conditioning federal funding — each of which takes time and can be contested.
- How current is this page, and how should I read the entries?
- The page is maintained and carries an “as of” date near the title. Each entry states its status — in force, proposed, pending, enacted, or rescinded — and links to its source. Where an item is noted as being confirmed, treat it as reported rather than settled until the page is updated.
How to cite this article
APA
Abdullahi, K. M. (2026, July 21). Federal AI Governance — A Maintained Timeline. Techné AI. https://techne.ai/insights/federal-ai-governance-timeline
MLA
Abdullahi, Khullani M. "Federal AI Governance — A Maintained Timeline." Techné AI, July 21, 2026, https://techne.ai/insights/federal-ai-governance-timeline.
Plain text
Abdullahi, Khullani M. "Federal AI Governance — A Maintained Timeline." Techné AI, July 21, 2026. Available at: https://techne.ai/insights/federal-ai-governance-timeline
Get the next piece
The Techné Institute is the firm’s periodical on AI, work, and the law — Adverse Impact weekly and Duty of Care monthly, written for boards, GCs, and the advisors who serve them.
About the author
Khullani M. Abdullahi, JD, is an AI governance and compliance consultant and the founder of Techné AI, an independent advisory firm based in Chicago. She submitted written testimony to the Illinois Senate Executive Subcommittee on AI and Social Media; the substance of one of her recommendations was incorporated into an AI-risk impact study bill. She authored the AI Governance & D&O Liability briefing now in active circulation among practitioners and underwriters, maintains the Illinois AI Legislative Ecosystem tracker, and hosts the AI in Chicago podcast. Techné AI is an advisory firm, not a law firm.