Guest post by Tommaso Del Becaro
The publication of the long-awaited America’s AI Action Plan by the federal government of the United States in July 2025 represents a crucial juncture in AI governance. Aimed at maintaining the US’ “unquestioned and unchallenged global technological dominance”, the AI Action Plan is the latest step by the Trump administration towards the deregulation of the AI sector and an even closer dialogue between public and private actors, particularly in the defence and security context.
The beginning of Trump’s second presidential term has been immediately characterized by a strong interest in emerging technologies, particularly AI systems. Trump’s first administration had already carried out various initiatives in this sector. Among these was the Summit on Artificial Intelligence for American Industry, hosted in 2018 at the White House, during which then Assistant to the President for Technology Policy, Michael Kratsios, remarked on the importance of US global leadership in this field. On that occasion, Kratsios stressed the need to “remov[e] barriers to innovation wherever and whenever we can to let American industry, American thinkers, and American workers reach their greatest potential”. The thinking that emerged from this Summit appeared to influence the entire US AI strategy over the following years, with the concept of “removing barriers” coming back into vogue immediately after the start of Trump’s second mandate.
The status held by the US worldwide in terms of AI development means that Washington’s approach to AI governance is likely reshaping the future of AI technologies, including their military applications. Public policies and statements from top governmental and industry officials reveal shifting dynamics in AI governance spanning the end of the Biden administration and the beginning of Trump’s second term. In this context, the AI Action Plan serves as a crucial lens through which to assess the implications for the private sector and AI firms’ decisions to profit from the militarization of these systems.
Removing Barriers: The New Trump Presidency’s Approach towards AI Governance
Compared to the previous Biden administration, the changes in US AI governance under Trump’s 2.0 presidency, as suggested by Brianna Rosen, could be evaluated in evolutionary rather than revolutionary terms. Despite the disruptive rhetoric maintained by the new president, both administrations have emphasized the necessity of developing a thriving domestic industry in the AI sector. In fact, the two approaches share a focus on the adoption and exploitation of AI systems through the legislation capable of accommodating their energy and computational demands.
To these elements of similarity is added the change of position of the Democratic administration toward the securitization of AI in 2024, thus paving the way for an increasingly greater presence of AI companies in the military sector. Signaling this transition was the publication of the Memorandum of 24 October 2024, which emphasizes the need to exploit AI systems within the national security strategy. There, AI is defined “as an era-defining technology” which “has demonstrated significant and growing relevance to national security”. For these reasons, “[T]he United States must lead the world in the responsible application of AI to appropriate national security functions”.
Considering this, Biden’s administration displayed a policy-making attitude that the AI Now Institute characterized as “far-from-coherent“, taking sometimes contradictory positions and clearly moving away from a previous focus on safety and transparency in the last year of the presidency. Notwithstanding this shift, Biden’s government had nonetheless brought significant changes in AI regulation, for example by extending pre-existing provisions to the private sector. Indeed, Biden used the occasion of his farewell speech to attack that sector, warning against a tech-industrial complex that is taking on the characteristics of an oligarchy in the US. This perspective was arguably further reinforced following the evident rapprochement of some representatives of such companies to the incoming Trump administration.
This proximity was then confirmed from the very beginning of the second Trump administration, with major representatives of tech companies in the front row at the President’s inaugural ceremony, followed by the announcement of billion-dollar projects on AI infrastructures. Trump’s 2.0 presidency changed the whole US legislative framework on AI within days of returning to the White House in January 2025, continuing prioritization of AI and effectively making his second mandate also an AI race mandate.
This happened by enacting, among others, Executive Order 14179 “Removing Barriers to American Leadership in Artificial Intelligence”, revoking “certain persisting AI policies and directives that act as barriers to American AI innovation”, and setting the goal of developing an Action Plan on AI within 180 days. Added to these initiatives was the proposed, later rejected, insertion of a ten-year moratorium in the One Big Beautiful Bill Act that would block State initiatives aimed at AI regulation, indicative of the federal government’s ambitions.
During the period between the publication of the Executive Order and that of America’s AI Action Plan, statements by major US government representatives had made clear Washington’s position regarding the regulation of AI technologies. During the Paris AI Action Summit in February 2025, US Vice President J.D. Vance stated a new industrial revolution “will never come to pass if overregulation deters innovators from taking the risks necessary to advance the ball, nor will it occur if we allow AI to become dominated by massive players looking to use the tech to censor or control users’ thoughts”. At the same time, Vance emphasized how “the AI future is not going to be won by hand-wringing about safety”.
This pro-innovation and anti-regulatory position was further reinforced in the months since January 2025, as the government published documents which underline the urgency of developing emerging technologies to consolidate Washington’s power globally and defending its national security. In that regard, the US explicitly speaks of the need to avoid over-regulation within Washington’s views on the application of AI in the military domain to General Assembly Resolution 79/239. There, the US underlines the need to avoid creating “new political slogans like ‘meaningful human control’ or developing new legal prohibitions or restrictions”; instead favoring the adoption of concrete measures “to harness the opportunities for AI in the military domain”.
America’s AI (Race) Action Plan
The US’ ambitions were validated with the publication of the long-awaited Action Plan. The document, divided into three pillars, focuses on:
- Accelerating AI Innovation;
- Building American AI Infrastructure;
- Leading in International AI Diplomacy and Security.
Within it, the primary objective emerges immediately: “Winning the Race”, thus maintaining the sense of urgency and competition typical of the US narrative on AI systems: “The AI race is America’s to win, and this Action Plan is our roadmap to victory“. The plan repeatedly emphasizes the dangers posed by the People’s Republic of China’s approach to AI, underlining the need to counter Beijing’s influence in international governance bodies. The document goes beyond building a national AI strategy, proposing itself as a “national security imperative, a geopolitical contest, and a civilizational opportunity”.
AI Patriots: Silicon Valley’s American Pride
Trump’s AI goals intertwine with the traditional blend of public and private actors in the US, as certified by the numerous meetings between the President and the heads of AI firms. Emblematic is the case of the dinner held on 4 September 2025 at the White House, where the CEOs of Big Tech confirmed their enthusiasm regarding the steps undertaken by the US administration. Among the interventions, the one delivered by Oracle’s CEO, Safra Catz, was particularly notable, emphasizing how President Trump has “unleashed American innovation and creativity” and that his work “is making it possible for America to win”.
Dispelling any doubt, the direction taken with the recent Action Plan is explicitly that of a private sector that “must be unencumbered by bureaucratic red tape”, with the White House itself underlining how the document has been acclaimed positively across the technology industry. On the day of the Action Plan’s publication, Trump defined the “red tape” as capable of strangling innovators, who must return to being “awarded with green light”. The document, as Christoph Müller observed, “contains almost no provisions which challenge the power of the tech giants”. For that reason, according to critics, it represents a substantial giveaway to Big Tech.
A further source of interest is represented by the “winning the race” narrative itself. Private companies, in fact, frame AI development through this interpretative lens, viewing national security issues as central. Not surprisingly, already starting from Trump’s re-election in November 2024, the defence tech sector has experienced an exponential increase in activity, with AI companies progressively “loosening the bounds”, carving out an increasingly important role as suppliers of AI-based products for defence and security to the Pentagon.
The Action Plan dedicates a paragraph to promoting AI adoption within the Department of Defense (DoD), recognizing its transformative potential in both combat and administrative operations, and clearly favoring this trend. The Plan envisions that “the United States must aggressively adopt AI within its Armed Forces if it is to maintain its global military preeminence while also ensuring […] its use of AI is secure and reliable”. This point was also iterated by Secretary of War Pete Hegseth before US military forces on 30 September 2025 at Quantico, emphasizing the need for “more AI in everything and ahead of the curve”. The need for “winning the race” in this sense reinforces the fusion between economic and strategic interests, representing a critical asset for US AI firms, spanning from Big Tech and tech startups.
This tendency is confirmed by the growing influence of such companies in the US military apparatus, with these same companies selling their quasi-messianic narratives. The case of Palantir Technologies illustrates this well, with CEO Alex Karp repeatedly supporting the rhetoric of an AI race against China. In Palantir’s view, as explained by Vanderborght and Nadibaidze, AI systems represent the key to safeguarding democracy and Western values. The firm publicly supported the new administration’s AI stance for months, contributing through its comments to the development of the Action Plan itself. Once published, it reiterated its pride in supporting a document capable of representing “the written source code for the next American century”.
Conclusions
The promise to liberalize investments and the “try-first” approach, endorsed in the Action Plan, opens enormous profits for AI firms, pushing for greater investments in the sector. The result is a Silicon Valley where, under Trump 2.0, a patriotic sentiment seems to resurface, with the defence sector ready to represent a significant portion of revenues, by virtue of the narrative that prioritizes winning the global AI race. Trump’s victory brought with it a substantial rebranding of military contracting, transformed for these companies into “a proud national duty”, with the public-private partnership on AI constituting the cornerstone of the US national security strategy for the near future.
About the author
Tommaso Del Becaro is a PhD student in Social Sciences at University of Genoa. His research focuses on AI governance in the EU context and its impact on Brussels’ digital sovereignty ambitions, with particular attention to transatlantic relations during Trump’s second term. He holds a Master’s degree in International Studies and a Bachelor’s degree in Political Science from the University of Pisa. His research experience includes a period at the University of Southern Denmark, where he was an AutoNorms guest researcher under Professor Ingvild Bode’s supervision.
