The AI executive order most people are asking about is Executive Order 14110, signed by President Biden on October 30, 2023, and revoked by Executive Order 14148 on January 20, 2025.1Federal Register. Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence For roughly fifteen months it was the most sweeping federal attempt to regulate artificial intelligence in U.S. history. It is no longer in force. The mandates it created on AI developers and federal agencies ended the day it was revoked, though several programs and rules launched under it survived because they rest on separate legal authority.
Is the AI Executive Order Still in Effect
No. Executive Order 14148, titled “Initial Rescissions of Harmful Executive Orders and Actions,” revoked EO 14110 in its entirety on January 20, 2025, along with dozens of other Biden-era executive actions. The order directed the White House Domestic Policy Council and National Economic Council to review federal actions taken under the revoked orders and rescind, replace, or amend them.
As of 2026, no replacement AI executive order with comparable scope has been issued. That leaves federal AI oversight without a central executive directive. Individual agencies still have whatever authority their statutes give them, but the coordinating mechanism the order created, with deadlines, reporting duties, and cross-agency alignment, is gone.
What EO 14110 Required While It Was Active
The most concrete obligation the order placed on private companies came through the Defense Production Act. Firms training a dual-use foundation model above a computing threshold of roughly 10 to the 26th power floating-point operations had to notify the Department of Commerce during development and share the results of their safety testing.2GovInfo. 3 CFR 14110 – Executive Order 14110 of October 30, 2023 The Defense Production Act backs its reporting mandates with criminal penalties of up to $10,000 in fines and one year of imprisonment for noncompliance.3Federal Emergency Management Agency. Defense Production Act of 1950
Companies also had to share results from red-team safety tests, meaning adversarial evaluations designed to make a model produce dangerous outputs such as instructions for biological or chemical weapons, or exploitable cybersecurity vulnerabilities. The point was to catch those failures before deployment. All of that is gone. No federal mandate currently requires AI developers to report training runs or turn over red-team findings to any agency.
Beyond the developer reporting piece, the order directed federal agencies to address algorithmic bias in sentencing tools, tenant screening, and healthcare diagnostics; to invest in privacy-preserving technologies; to examine how agencies purchase personal data from commercial brokers; and to develop labor principles covering AI-driven workplace surveillance and job displacement. Each of those directives lost its legal force with the revocation.
What Survived the Revocation
Some work initiated under EO 14110 continues, because it was authorized by separate legislation, funding, or formal rulemaking rather than resting on the order alone.
NIST Frameworks
The National Institute of Standards and Technology’s AI Risk Management Framework remains active as a voluntary tool. It organizes risk management around four functions: govern, map, measure, and manage.4National Institute of Standards and Technology. AI Risk Management Framework In July 2024, NIST released a companion profile for generative AI, the AI 600-1, focused on organizational governance, content provenance, pre-deployment testing, and incident disclosure after release.5National Institute of Standards and Technology. Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile Because these are voluntary, they were not affected by the revocation. Many companies still reference them when building internal governance programs, and they inform state legislation now moving through statehouses.
NIST’s watermarking research, aimed at distinguishing AI-generated content from human-created media, continues under the agency’s broader standards mission. No federal rule requires companies to implement watermarking on AI outputs.
H-1B Modernization
The order directed agencies to streamline visa processing for AI researchers, focusing on H-1B specialty worker visas and O-1A extraordinary ability visas.6Federation of American Scientists. Unlocking American Competitiveness: Understanding the Reshaped Visa Policies under the AI Executive Order In December 2024, the Department of Homeland Security published a final rule modernizing H-1B requirements, effective January 17, 2025. The rule makes it easier for H-1B holders to move between jobs and improves access for entrepreneurs.7Federal Register. Modernizing H-1B Requirements, Providing Flexibility in the F-1 Program, and Program Improvements Affecting Other Nonimmigrant Workers Because it went through notice-and-comment rulemaking, it did not vanish when EO 14110 did.
National AI Research Resource
The National AI Research Resource, administered by the National Science Foundation, gives researchers, startups, and small businesses access to computing power and datasets that would otherwise sit only inside large tech companies. The program has supported more than 600 research projects and 6,000 students across all 50 states, Washington, D.C., and Puerto Rico.8U.S. National Science Foundation. National Artificial Intelligence Research Resource It has its own authorization and funding, so it continues to operate.
What Still Applies to AI Without the Order
Federal civil rights laws that prohibit discrimination in housing, lending, healthcare, and employment predate EO 14110 and still apply to algorithmic decision-making. Agencies retain authority to enforce those statutes against biased AI systems even without an executive mandate telling them to focus on it. Fourth Amendment questions about the government buying personal data from commercial brokers remain unresolved and continue to draw attention from Congress and oversight bodies, though no executive directive currently compels agencies to justify those purchases.
On labor, no comprehensive federal law restricts specific forms of automated workplace surveillance, though several states have introduced bills addressing the issue.
Where That Leaves Companies and Workers in 2026
For a company building or deploying AI, federal regulation is now a patchwork rather than a program. Existing statutes covering discrimination, consumer protection, workplace safety, and sector-specific practices apply to AI the same way they apply to any other tool. Voluntary NIST frameworks fill in where binding rules do not, and state legislation is moving faster than anything in Washington.
For workers and consumers, the practical picture is similar. The rights that existed before the order still exist. What is missing is the unified federal push that briefly asked every relevant agency to treat AI governance as a shared priority, on the same schedule, under the same standards. Whether Congress or a future executive order restores that coordinating layer is the open question shaping AI policy going forward.