What:	Signing of Executive Order 13859
When:	February 11, 2019
Why it matters:	The document — titled “Maintaining American Leadership in Artificial Intelligence — was the first wide-ranging executive order on U.S. AI governance. But it set the tone for a White House-led approach to AI, and the five other EOs that followed.

To celebrate our 25th anniversary in 2025, we’re spotlighting 25 pivotal moments that reshaped the way the government operates. Visit our anniversary page each week as we look back on these defining moments.

Across three White House administrations, there have been six executive orders on artificial intelligence in just as many years — a succession of presidential actions that reflects both a quickly changing technology landscape, but also shifts in the nation’s overall approach to AI governance in the absence of major legislation on the topic.

But it was Executive Order 13859, signed in February 2019, that started the trend toward regulating AI from the White House. That order, signed during President Donald Trump’s first administration, directed federal agencies to prioritize AI research and development (R&D) and workforce development, aimed to make federal data and models available for AI development work, told agencies to create guidance for the use of AI in the industries they regulate and called for an action plan to protect the U.S.’s technological advantage in AI.

“It was some of the earliest activity directly addressing artificial intelligence technology, and this was obviously well before the ChatGPT era that we’ve seen now,” Neil Chilson, the head of AI policy at the Abundance Institute, told Federal News Network in a recent interview. “Good governance is good governance. You want the tools that work, that help you achieve your job … and those strengths and limits continue to change as the technology itself continues to improve. And so I think people who are using this in the federal government have a tough challenge, but it’s also a really exciting one. These are powerful tools, and using them properly is going to be really important.”

But in many ways, that first order was a reflection of AI work that was already ongoing in federal agencies. Within days of its signing, the Defense Department, for instance, formally unveiled its then-new Joint Artificial Intelligence Center and tasked it with finding ways to use AI for both warfighting applications and back-office functions. The JAIC became the successor to Project Maven, the department’s AI task force that had already been up and running for the past two years. The goal, in part, was to build a base of AI expertise and then spread it throughout the military services.

“It’s a real simple concept: Bring people in, give them a place to work and learn over the next two years and actually send them back out to their respective services with an enormous amount of talent and capability and know-how of what JAIC can do,” Dana Deasy, DoD’s then-chief information officer, told reporters at the time. “It’s going to be made up of civilians, academics, the commercial world and the service side. We think that service side provides that linkage.”

Trustworthy AI

But even as agencies moved to carry out that first order, its implementation also laid the groundwork for a second Trump executive order that was more directly targeted at the activities of federal agencies themselves.

That order, at the tail end of the first Trump administration in December 2020, aimed to create a framework for federal agencies’ “trustworthy” use of AI in their own missions, while also adding to the ranks of AI experts within those agencies, including by creating a specialized AI track within the Presidential Innovation Fellows program.

“We will continue to develop new technologies in a way that advances innovation, promotes public trust, protects civil liberties, and remains consistent with our common principles,” Michael Kratsios, the former federal chief technology officer, said in remarks ahead of the EO’s signing. “Our holistic strategy will improve our development of AI, empower the American people, promote innovative uses of new technology and stay true to our values. We start from a position of great strength, and we have a plan to keep winning.”

After a change of administrations the following month, it took the Biden administration three more years to issue its first AI executive order. But when that White House did put its mark on federal AI policy, the order, with 13 separate sections, was widely seen as the U.S. government’s most far-reaching attempt at AI governance to date.

It did echo many of the previous themes, however, including a major focus on developing the AI workforce, including a “governmentwide AI talent surge” and a major emphasis on safeguards to put ethical boundaries around AI use and ensure the technology’s trustworthiness. It also called for the creation of the first-ever chief federal AI officer and the development of a new strategy for the public’s use of AI.

“The executive order is about AI safety and security,” Biden said in 2023 remarks during the order’s signing. “Today, I am invoking what’s called the Defense Production Act that federal government uses in the most urgent of moments, like mobilizing the nation during — a nation in time of war or developing COVID vaccines during the pandemic. This executive order will use the same authority to make companies prove — prove that their most powerful systems are safe before allowing them to be used. I want to — before allowing them to be used. That means companies must tell the government about the large-scale AI systems they’re developing and share rigorous independent test results to prove they pose no national security or safety risk to the American people.”

Consumer-facing AI

And the timing of the order was important, with consumer-facing use of AI technologies beginning to burst onto the scene in major ways that year, Jonathan Alboum, the former chief information officer at the Agriculture Department, told Federal News Network at the time.

“The arrival of high-quality, consumer-facing generative AI made an impact in 2023 on par with the launch of the iPhone in 2007. As commercial organizations integrate GenAI tools into their operations, there will be an expectation by the public that government does the same,” he said. “However, the stakes for government are much higher, making adoption a challenge in 2024 and beyond. GenAI tools built on general purpose large language models (LLMs) pose the risk of producing inaccurate or biased information, which is unacceptable in a public setting.”

Like the first Trump administration, the Biden White House followed up its first order with a second during its final days in office. That order, signed in January of this year, focused on the physical infrastructure needed to build out AI capabilities across the country, including by making federal sites available for data centers and the power generation facilities needed to run them.

“We will not let America be out-built when it comes to the technology that will define the future, nor should we sacrifice critical environmental standards and our shared efforts to protect clean air and clean water,” Biden said in signing his second order.

Second Trump administration and new action plan

But the Biden White House’s AI orders were short-lived — on paper anyway. As one of his first acts as president, amid a flurry of other executive orders, President Trump revoked Biden’s landmark 2023 AI order and replaced it with a brief new one of his own.

The order, on its own, created no new substantive policy, but instructed White House aides to develop an Artificial Intelligence Action Plan within the administration’s first 180 days. That plan is still in development after the Office of Science and Technology Policy spent the first two months of the administration accepting public comments.

Additionally, in revoking the Biden orders, Trump said he was eliminating elements of the AI governance agenda that his administration saw as overly prescriptive or could lead to “ideological bias or engineered social agendas.”

“With the right government policies, we can solidify our position as the global leader in AI and secure a brighter future for all Americans,” he wrote. “This order revokes certain existing AI policies and directives that act as barriers to American AI innovation, clearing a path for the United States to act decisively to retain global leadership in artificial intelligence.”

While the action plan is still in development, the administration did begin to put some details around its revised AI agenda in April, when the Office of Management and Budget issued a memo telling agencies to focus their efforts on “high impact” use cases, and the president signed another order focusing on AI in education — both training students on AI technologies and using it in classrooms.

“By establishing a strong framework that integrates early student exposure with comprehensive teacher training and other resources for workforce development, we can ensure that every American has the opportunity to learn about AI from the earliest stages of their educational journey through postsecondary education, fostering a culture of innovation and critical thinking that will solidify our nation’s leadership in the AI-driven future,” Trump wrote.

Role of Congress in AI leadership

Meanwhile, Congress has largely taken a back seat to the three administrations when it comes to AI policy development.

Lawmakers, as part of the 2021 Defense authorization bill, did pass the National Artificial Intelligence Initiative, which created the first statutory definition of AI, created a new AI office and task force led by OSTP and the National Science Foundation and stood up an interagency committee to coordinate federal AI programs.

On Capitol Hill, lawmakers have been active with an AI task force of their own, and it’s produced sweeping proposed bipartisan legislation, but the 250-page bill has yet to make it to a vote.

Rep. Don Beyer (D-Va.), a member of the 24-member task force, said the legislation — endorsed by all 24 members — would codify many areas of consensus across the Trump and Biden presidencies, and also catch the U.S. up to the rest of the world in creating a statutory framework for AI governance, rather than ones established by executive orders and changeable by future ones.

“One of the main areas of difference, and it’s not profound, is how much regulation should there be,” Beyer told Federal News Network. “The European Union has passed its own EU AI Act, which is widely seen around the world as overly prescriptive and restrictive. Lots of licensing requirements and at least the business community feels strongly that this will dampen any creativity, any real initiative. Our committee has said we want regulation, but we want a light touch. We want to make sure that American entrepreneurs and the creative young people are doing their best without government getting in their way. But we also don’t want the Wild West. We don’t want terrorists, organized crime, child sexual assault, pornographers to just have free range without any regulation. That’s not acceptable.”

The post 2019 executive order began a trend toward White House-centered AI policy first appeared on Federal News Network.

Vicky