In Washington, rare bipartisan agreement exists on one point: artificial intelligence will shape — indeed, decide — who wins the 21st century. But what it means to “win” divides President Donald Trump from his predecessor, Joseph Biden.
The Trump Administration’s new AI Action Plan marks a sharp reversal in approach to AI governance, industrial policy, and national security. While Biden aimed to place guardrails around a fast-moving and unpredictable technology, Trump is bullish and unapologetically pro-business. His Action Plan comprises 90 recommendations, underpinned by three executive orders. It calls for slashing environmental regulations, fast-tracking data center construction, preempting restrictive state laws, and promoting AI exports. The aim, in Trump’s words, “America started the AI race. And we’re going to win it.”
By abandoning AI safeguards, the US will pile pressure on Europe — and perhaps Silicon Valley as well. Although Europe is considering a pause on its AI Act, the continent remains concerned about minimizing the potential risks associated with the new technology. Will it resist US deregulation, even at the possible cost of losing the AI race? US companies, already under scrutiny in Europe, may find it challenging to navigate contradictory policies on both sides of the Atlantic.
Under Biden, Washington shared many European concerns. It treated AI as a dual-use technology — full of promise, while fraught with risk. The Biden administration imposed controls on chip exports to China, required AI developers to disclose safety data, and directed federal agencies to ensure their systems were free from bias and discrimination.
Trump has dismantled this policy. Gone are the “diffusion rules” that limited AI computing power to certain countries, as are procurement safeguards, civil rights audits, and transparency requirements.
Gone, too, is Biden’s high-fence approach, which sought to restrict advanced AI capabilities for national security reasons. It is the “full-stack export package” — a strategy for bundling AI hardware, software, and American technical standards into cohesive offerings for countries with substantial budgets. The Trump White House prefers out-exporting rather than out-regulating China. Case in point: lifting the NVIDIA H20 chip exports to China.
Another dumped Biden priority: environmental regulation. The Trump AI plan proposes carving out exemptions to accelerate data center construction and other related power infrastructure.
In the Trump vision, the federal government’s word should be supreme, overruling any state AI laws that could “strangle innovation.” In theory, this could simplify compliance. In practice, it centralizes control over AI governance in the executive branch.
While deregulating, the Trump administration wants to shape the way AI presents information. Conservatives accuse AI models of a liberal bias. In response, Trump intends to purge what he calls “woke AI.” An Executive Order mandates that all large language models used by federal agencies must be “ideologically neutral.”
Trump’s AI strategy is, in many ways, a natural extension of his economic worldview: deregulatory, export-oriented, and pro-free markets. It reflects a particular understanding of the tech industry’s role in global power. While the Biden administration tried to balance AI development with safety, rights, and democratic oversight, the Trump White House seems to view such constraints as indulgences.
The Trump plan does devote attention to a different concern: security by design. It highlights the vulnerabilities of AI systems to data poisoning, input manipulation, and privacy breaches that could undermine their performance or reliability. The plan proposes a coordinated effort led by the Department of Defense, in partnership with NIST, the Department of Commerce, and the Office of the Director of National Intelligence, to refine and institutionalize AI assurance standards, frameworks, and toolkits.
With its AI Action Plan, the Trump administration has given American big tech nearly everything it wanted: deregulation, export support, and permissive rules for training data. Whether that bargain produces a technological renaissance or a regulatory reckoning remains to be seen.
Elly Rostoum is a Google Public Policy Fellow with the Center for European Policy Analysis (CEPA). She is a Lecturer at Johns Hopkins University. You can find out more about her work here: www.EllyRostoum.com
Bandwidth is CEPA’s online journal dedicated to advancing transatlantic cooperation on tech policy. All opinions expressed on Bandwidth are those of the author alone and may not represent those of the institutions they represent or the Center for European Policy Analysis. CEPA maintains a strict intellectual independence policy across all its projects and publications.
2025 CEPA Forum Tech & Security Conference
Explore the latest from the conference.
