Executive Order 14110, formally titled "Executive Order on Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence" (often shortened to "Executive Order on Artificial Intelligence"), was the 126th executive order signed by U.S. President Joe Biden on October 30, 2023. It defined the administration's policy goals for artificial intelligence (AI), directing executive agencies to take specific actions. The order was widely considered the most comprehensive piece of U.S. government governance on AI at the time of its issuance. It was rescinded by President Donald Trump on January 20, 2025, within hours of his assuming office.
The order emerged amid rapid advancements in generative AI technologies, including the widespread deployment of large language models such as ChatGPT by OpenAI. It sought to balance innovation with risk mitigation, addressing concerns ranging from national security threats to civil liberties and market competition.
Background
The drafting of Executive Order 14110 was motivated by the rapid development of generative AI models in the early 2020s, particularly following the public release of ChatGPT in November 2022. It was the third U.S. executive order focused on AI, following two earlier orders signed by former President Donald Trump. Experts and policymakers raised concerns about AI's potential to spread misinformation, exacerbate discrimination, and pose long-term existential risks.
In August 2023, Arati Prabhakar, director of the White House Office of Science and Technology Policy (OSTP), indicated that the administration was expediting work on AI governance. A week before the order was issued, Prabhakar stated that guidance from the Office of Management and Budget (OMB) would follow soon after. The order was formally signed by President Joe Biden on October 30, 2023, as Executive Order 14110.
Key provisions
The order outlined policy goals including promoting competition and innovation in the AI industry, upholding civil and labor rights, protecting consumers and privacy from AI-enabled harms, developing watermarking systems for AI-generated content, and maintaining U.S. global leadership in AI. It also directed federal agencies to adopt procurement and use policies for AI, and to coordinate with private entities on AI security matters.
One of the most notable structural requirements was that major federal agencies establish dedicated chief artificial intelligence officer positions. The legislation did not create new federal agencies but rather mandated coordination across existing bodies such as the Department of Homeland Security (DHS), Department of Commerce's National Institute of Standards and Technology (NIST), and the Department of Veterans Affairs.
Background
The order was drafted in response to rapid advances in Generative AI and the release of ChatGPT, a Large language model. Prior to 2023, the Biden administration had published a Blueprint for an AI Bill of Rights and secured non-binding safety commitments from major technology companies. Concerns raised by experts included risks of misinformation, discrimination, and potential existential threats from advanced systems.
In August 2023, Arati Prabhakar, director of the White House Office of Science and Technology Policy, stated that work on the order was being accelerated. Office of Management and Budget (OMB) guidance was expected to follow the order's issuance.
Policy Goals and Provisions
The order sought to balance innovation with risk mitigation. Its stated goals included promoting competition in the AI industry, preventing AI-enabled threats to civil liberties and national security, and ensuring U.S. global competitiveness in the field. It directed agencies to develop watermarking systems for AI-generated contentColorado and to address intellectual property concerns related to generative models.
The order required numerous major federal agencies to create dedicated chief artificial intelligence officer positions. Following the signing, the General Services Administration (GSA), the Department of Education, and the National Science Foundation (NSF) confirmed appointments or elevations of officials to these roles. The order also specified agency responsibilities: the Department of Homeland Security (DHS) was tasked with developing AI-related security guidelines, including for critical infrastructure sectors like energy and cybersecurity; the Department of Veterans Affairs was directed to launch an AI competition for reducing healthcare worker burnout; and the National Institute of Standards and Technology (NIST) was to develop a generative AI resource to supplement its AI Risk Management Framework.
Policy Goals and Provisions
The order articulated several high-level goals: promoting competition and innovation in the AI industry; upholding civil and labor rights; protecting consumers and their privacy; specifying federal procurement and use policies; developing watermarking systems for AI-generated content; and maintaining U.S. global leadership in AI. It mandated actions across multiple federal agencies, including the creation of dedicated chief artificial intelligence officer positions within major departments.
A central theme was the balance between capturing AI's economic and societal benefits and mitigating risks. Biden stated at the signing that AI was being developed at "warp speed," and that governance was needed to "realize the promise of AI and avoid the risk." The order also emphasized the development of watermarking for AI-generated content and protections against intellectual property theft from generative models.
Impact on Agencies
Creation of chief AI officer positions
The order required many large federal agencies to appoint a chief artificial intelligence officer (CAIO). Within days of signing, the General Services Administration (GSA) and the U.S. Department of Education confirmed appointments, and the National Science Foundation (NSF) elevated an existing official to the role. Other agencies followed suit, formalizing AI leadership across the federal government.
Department responsibilities
The Department of Homeland Security (DHS) was tasked with developing AI-related security guidelines, including cybersecurity measures undersized and coordination with private sector firms in critical infrastructure such as energy. The Department of Veterans Affairs was directed to launch an AI technology competition aimed at reducing occupational burnout among healthcare workers through AI-assisted tools for routine tasks. The National Institute of Standards and Technology (NIST) was instructed to develop a generative-AI-specific resource to supplement its existing AI Risk Management Framework.
Analysis and scope
The executive order was notable for its broad scope but also had defined limits. It did not include proposals for a licensing regime for advanced AI models, an idea supported by some industry leaders. It also did not prohibit high-risk AI uses or mandate that companies disclose training data or model details. The order instead relied mainly on federal agency directives and voluntary industry commitments.
The order required departments such as the Department of Veterans Affairs to launch an AI competition to reduce healthcare worker burnout)SkipIncorrect mid-sentence, correct the mistake, and continue. (Note: the sentence should not contain the word 'SkipIncorrect'.)
and tasked the Department of Commerce's NIST with developing a generative AI-focused supplement to its existing AI Risk Management Framework. The Department of Homeland Security was directed to create AI security guidelines and coordinate with private firms in critical infrastructure sectors.
Agency implementation
In the days after the order, the General Services Administration and the U.S. Department of Education confirmed the appointment of chief AI officers Bonny Simi and Kevin Mills respectively. The National Science Foundation also elevated an official to the chief AI officer role. The order required agencies to integrate AI governance into their operations, though it did not mandate specific hiring levels or budgets.
Response and Reception
Praise came from many Democratic lawmakers. Senator Richard Blumenthal (D-CT) and Representative Ted Lieu (D-CA) voiced support. Representative Don Beyer (D-VA), chair of the House AI Caucus, called it a "comprehensive strategy for responsible innovation" while urging Congress to pass legislation. The editorial board of the Houston Chronicle described it as a "first step toward protecting humanity."
Republican Senator Ted Cruz criticized the order, saying it created "barriers to innovation disguised as safety measures." The Chamber of Commerce and tech industry groups including NetChoice and the Software and Information Industry Association argued it could hinder private-sector innovation. Public polling by the AI Policy Institute found 69% support overall, with 78% among Democrats, 64% among Republicans, and 65% among independents. Some civil society leaders praised its provisions on "AI fairness," and Democratic representatives including Blumenthal and Lieu endorsed the order.
Implementation and Impact
The order required several federal agencies to issue new rules and guidance. Under its mandate, NIST developed a framework for generative AI risk management. The OMB was tasked with coordinating agency actions aides, though specific timelines were not fully specified. The order did not include proposals for a licensing regime for advanced AI models, nor did it mandate disclosure of training data or prohibit specific high-risk uses of AI. These omissions were noted by commentators at Axios, who observed the order's scope was broad but not exhaustive.
Reception and Rescission
Reception was sharply divided. Editorial boards and Democratic lawmakers, including Representative Ted Lieu (D-CA) and Senator Richard Blumenthal (D-CT), praised the order. Representative Don Beyer (D-VA), chair of the House AI Caucus, called it a "comprehensive strategy for responsible innovation." Republican Senator Ted Cruz criticized it as creating "barriers to innovation disguised as safety measures."
Industry groups, including the Chamber of Commerce, NetChoice, and the Software and Information Industry Association, argued the order could hamper innovation. Civil society organizations largely welcomed its focus on fairness and risk mitigation. A poll by the AI Policy Institute reported that 69% of voters supported the order, with 78% support among Democrats, 65% among independents, and 64% among Republicans.
Legacy
Executive Order 14110 was rescinded on January 20, 2025, when President Trump signed an executive order revoking several Biden-era policies shortly after taking office. The rescission nullified its mandates, though many agencies had already implemented structures such as chief AI officer roles. The order remains a reference point for subsequent federal AI governance discussions.