Last week, the White House released its long-awaited AI Action Plan, setting out steps to reshape the U.S. approach to artificial intelligence. The plan includes a number of welcome provisions—particularly its emphasis on innovation and deployment—but its overall focus leans more toward industrial strategy than well-developed regulatory design. For those who favor an innovation-friendly, proportionate approach to AI, the plan raises several important questions. Its long-term significance will depend not only on how its proposals are implemented, but also on whether future efforts address persistent challenges such as policy volatility and growing misalignment between U.S. diplomacy and tech policy objectives.
The plan follows an earlier request for public input on the future direction of U.S. AI policy, to which NTUF submitted comments calling for a well-designed, proportionate, and innovation-focused regulatory approach. At 28 pages, it marks the first major AI policy document issued under the second Trump Administration and is structured around three pillars: accelerating AI innovation, building infrastructure, and advancing AI diplomacy and security. While each of these areas warrants closer scrutiny, this article focuses on overarching strengths and limitations rather than offering a proposal-by-proposal assessment.
Compared to the Biden Administration’s approach, the new plan places stronger emphasis on AI innovation and deployment, including proposals to identify regulatory barriers that may inhibit adoption. Among these is a directive for the Office of Science and Technology Policy to solicit evidence on regulations that impose disproportionate burdens. The plan also endorses tools such as regulatory sandboxes, though it offers little detail on how such mechanisms would be designed or implemented.
Beyond its emphasis on innovation, the remainder of the Action Plan outlines several priorities that align with the administration’s broader policy priorities. The second pillar includes proposals to expand the use of AI across federal agencies, improve public data infrastructure, strengthen incident response capacity, and address computing constraints, including the need for expanded data center infrastructure. It also emphasizes the importance of developing a more robust domestic AI workforce. The third pillar—centered on diplomacy—sets out objectives related to the export of U.S. AI products and services, with a particular focus on extending the global reach of U.S. standards and technical systems.
Yet these priorities are undercut by unresolved tensions and gaps in the broader U.S. approach to AI governance.
First, a major challenge for U.S. AI policy is its growing susceptibility to political volatility. While the Action Plan includes some recognizably bipartisan priorities—such as workforce development and efforts to address risks from frontier models and biosecurity—much of its strategy marks a clear departure from the approach pursued in recent years. While a shift toward a more market-oriented approach is welcome, the new plan deprioritizes key AI risk and safety frameworks developed in recent years.
The new strategy also omits the multilateral initiatives emphasized by the previous administration, including U.S. engagement through forums such as the OECD and the Global Partnership on AI. While diplomatic engagement in such institutions may not carry widespread domestic appeal, this turn away from collaborative norm-setting represents a missed opportunity for the United States to contribute meaningfully to international frameworks on responsible AI —while advancing a proportionate, innovation-friendly approach.
By contrast, in other G-7 countries—including those marked by political polarization—successive governments have generally maintained a degree of continuity in AI regulation across party lines. A similar approach in the United States would help reduce institutional uncertainty and ensure that progress is not reversed with each change in administration.
Second, while certain public investment in AI infrastructure is needed, the Action Plan’s overriding emphasis on industrial strategy risks drifting toward a more statist approach. From domestic chip manufacturing to next-generation data centers, the plan prioritizes subsidies, incentives, and public-private partnerships as central instruments of AI policy. Yet without well-designed safeguards, there is a growing risk that taxpayer resources will be used inefficiently—or that such interventions will reinforce politically favored sectors rather than strengthen long-term competitiveness.
Third, even in areas where the plan endorses promising ideas, it would benefit from a greater focus on concrete recommendations. References to “global AI dominance” and a coming technological “revolution” abound, while proposals such as regulatory sandboxes, open-source models, and public-sector adoption are mentioned with little detail on how they will be designed, governed, or evaluated. More clearly articulated implementation plans and governance and evaluation frameworks become especially important when substantial taxpayer resources are at stake; without them, even well-intentioned initiatives risk becoming fiscally inefficient exercises in political signaling.
It could be argued that an action plan is not the appropriate venue for detailed implementation. Yet the absence of even basic design principles—combined with the lack of a broader regulatory framework for AI and emerging technologies—remains a structural weakness. These gaps could be addressed through future executive action, but, for now, they leave the plan’s more ambitious proposals untethered from institutional reality.
Finally, the plan’s third pillar—focused on diplomacy—emerges as its weakest link. The Action Plan states that the United States must export its “full AI technology stack . . . to all countries willing to join America’s AI alliance,” warning that failure to do so would constitute “an unforced error.” Yet, by framing other countries primarily as markets to be captured in service of U.S. technological supremacy—as Kat Duffy of the Council on Foreign Relations has noted—the plan risks alienating both policymakers and publics, in advanced and emerging economies alike.
In the short term, a more pugnacious diplomatic approach may yield bilateral agreements that disproportionately advantage Washington, particularly when backed by tariff threats and asymmetric economic leverage. Yet, over time, it increases the likelihood that other countries will seek to reduce their technological dependence on the United States—just as the U.S. has sought to do in its bilateral relationship with China.
The plan also appears misaligned with the recent trajectory of U.S. foreign policy. As Rush Doshi, former China Director on the National Security Council, has noted, the push to promote U.S. standards globally sits uneasily with Washington’s reduced engagement in multilateral fora—including those increasingly central to AI governance. Whether the promotion of U.S. standards is ultimately beneficial depends on their substantive quality—but absent credible participation in international institutions, the United States’ ability to shape global norms is likely to diminish.
Taken together, the AI Action Plan reflects a renewed emphasis on innovation and deployment—but it would benefit from stronger institutional scaffolding to translate ambition into lasting impact. Absent a clearer framework for governance, implementation, and diplomatic engagement, even its most promising proposals risk underdelivering or backfiring. A more thoughtful and coordinated strategy—grounded in proportionate regulation, sound fiscal design, and credible diplomacy—will be essential not only to advancing a responsible U.S. approach to AI, but to doing so in a way that earns trust both at home and abroad.