The Trump administration will maintain the United States' strongly pro-innovation AI policy, and although the European Union may partially rethink its pro-regulatory position, divisions within the bloc will enforce continued provisions for AI safety. With a new administration in the United States and a new European Commission in the European Union, both jurisdictions are attempting to define their policies on artificial intelligence amid the global AI race, primarily by weakening AI regulations. For instance, U.S. President Donald Trump's administration is pledging to roll back the country's few AI rules and ensure that no new burdensome requirements stifle AI innovation. Meanwhile, some European leaders are raising concerns about Europe's ability to compete in AI development and rollout, as well as leverage AI to enhance economic competitiveness more broadly, amid the bloc's heavy regulatory environment. Thus, on both sides of the Atlantic, there appears to be a growing consensus that concerns over AI safety should take a back seat to staying competitive in the AI race.

• At the Feb. 10-11 AI Action Summit in Paris, France, U.S. Vice President JD Vance touted the U.S. desire for AI dominance and urged allies to follow in the United States' footsteps of prioritizing innovation over regulation. Private companies also advocated for acceleration, with Google CEO Sundar Pichai saying in a speech that "the biggest risk could be missing out." In addition, OpenAI executives told Axios that in private meetings at the summit, government leaders appeared to be more focused on aspirations to build massive data centers in their countries than AI safety concerns. 

The Trump administration says it is adopting a pro-innovation AI policy and vowed to remove allegedly burdensome Biden-era regulations that the White House claims stifle innovation and censor conservative views. Trump followed through with his campaign trail promise to repeal former President Joe Biden's AI executive order "on day one." Though the order largely contained voluntary commitments covering a fairly narrow set of issues rather than significant, widespread regulation, Trump and his allies referred to the Biden administration's approach to AI as "radical leftwing ideas" due to allegedly burdensome regulations that limited the AI sector's ability to advance innovation. On Jan. 23, Trump signed his own executive order entitled "Removing Barriers to American Leadership in Artificial Intelligence." Trump's order largely serves as a placeholder for more definitive initiatives, setting broad goals on the administration's overall intent for AI rather than laying out clear plans. These goals center on enhancing U.S. dominance in AI "to promote human flourishing, economic competitiveness, and national security." To this end, the order directs the development of an AI action plan within 180 days to help sustain U.S. dominance and states that the administration will continue to prioritize AI leadership and build on Trump's first term commitments to double AI research investment, as well as issue guidance for federal agencies' adoption of AI and provide guidance for AI development in the private sector. Trump's executive order also states that to maintain leadership in AI, the United States "must develop AI systems that are free from ideological biases or engineered with social agendas." Though the order does not provide specifics on this aspect, the clause is a nod to Republican criticism of AI chatbot guardrails and AI-enabled content moderation. Such guardrails and moderation efforts are intended to prevent the output of harmful, abusive and inaccurate/misleading information, but Trump and other prominent Republicans have strongly criticized these measures as infringing on free speech by allegedly only promoting liberal ideals and censoring conservative views. 

While framed as a stark departure from Biden's approach, Trump's AI policy largely aligns with the previous administration's stance, though chatbot guardrails and content moderation may weaken. Because the Biden administration had already taken a largely pro-innovation approach to AI with little actual regulation, Trump's policy to "deregulate" AI will have little practical impact in terms of advancing innovation. Also in alignment with Trump's plans, the Biden administration had expressed commitment to U.S. AI investment and development. Moreover, Biden's AI executive order mainly contained voluntary commitments like security testing, information sharing and transparency, and while such measures may be seen as burdensome, companies that adopted these guidelines may choose not roll back their existing practices due to shareholder pressures or concern about losing users. For example, shortly after Trump's election, spokespeople from Nvidia, Scale AI, Amazon, IBM, Salesforce and Inflection AI — which all signed onto Biden-era White House voluntary commitments — told FedScoop that their positions on safe AI had not changed, though many also emphasized the need for innovation and expressed their commitment to working with the new administration. Additionally, while federal AI legislation is unlikely to broadly regulate the private sector, individual states continue to introduce AI legislation, meaning compliance requirements in various jurisdictions may prompt companies to maintain safety standards to avoid potential state-level violations. That said, the Trump administration's anti-censorship position on issues like chatbot guardrails, content moderation and initiatives to protect against AI biases may change companies' actions in these areas. However, because binding legislation did not previously regulate these aspects of AI on a federal level, there would be no specific policy changes to remove legislation. Instead, the Trump administration could leverage enforcement agencies like the Federal Trade Commission to investigate chatbot guardrails, or the White House could put informal pressure on companies to roll back chatbot guardrails, discontinue use of AI-powered content moderation or remove commitments to protect against AI bias as part of a broader fight against DEI policies. Trump could also issue an executive order requiring tech companies to remove guardrails or change content moderation practices in order to keep existing government contracts or engage in public-private partnerships. 

• Former President Biden's initial AI executive order, signed on Oct. 30, 2023, included several provisions to enhance U.S. AI innovation, including launching a pilot of the National AI Research Resource, a push to hire data scientists and AI professionals across the federal government, ordering the National Science Foundation to fund and launch AI-focused NSF Engines and charging the NSF with prioritizing AI workforce development. Just before he left office, Biden also issued a Jan. 14 executive order on "Advancing United States Leadership in Artificial Intelligence Infrastructure," which the Trump administration has left in place. The order directs federal agencies to accelerate large-scale AI infrastructure development at government sites to ensure that such facilities can be built quickly and at scale to support advanced AI operations. 
• On Feb. 7, the White House said it is seeking public comment on priority actions to include in the AI Action Plan such as export controls, data centers, national security, data privacy, intellectual property and safety standards. The request mentions safety standards and overall risks, but it also emphasizes the administration's unwillingness to impose government requirements on the private sector. Though the government has not publicized the plan's details, it will likely contain guidelines, recommendations and actions to stifle Chinese AI development, rather than funding commitments, given the administration's drive to cut federal funding. It may also address the need for large-scale data centers to power AI, which Trump has acknowledged.
• California, which houses many of the country's leading tech companies, has been a leader in issuing regulations and in its most recent 2023-24 legislative session enacted 17 bills on AI that apply to specific sectors or issues, like privacy and healthcare. While Gov. Gavin Newsom vetoed California's highly anticipated AI safety bill that would have been much more stringent, he left the door open to future legislation. Additionally, Colorado has enacted a broad AI Act to address high-risk AI systems and impose transparency requirements for consumer-facing AI systems. 
• OpenAI announced on Feb. 12 updates to its Model Spec, a document that defines how the company wants its models to behave. OpenAI said the updated Model Spec "explicitly embraces intellectual freedom…no matter how challenging or controversial a topic may be," aligning with the Trump administration's position on chatbot guardrails. OpenAI's changes may have been an independent business decision, as the update will expand use cases for OpenAI's products, such as political analysis or more mature or creative screenwriting (despite genuine concern about lightened guardrails enabling a rise in harmful content, like hate speech). However, the timing of OpenAI's new Model Spec aligns with a broader trend of tech companies changing policies to curry favor with the new Trump administration, and the change in Model Spec supports conservative calls for such action. 

By contrast, the European Union leveraged its regulatory power to issue its AI Act, which has faced criticism amid a broader pushback in recent years against some of Brussels' more burdensome regulations. The EU AI Act, originally proposed in April 2021, officially went into effect in August 2024 after much deliberation on a revised draft released in June 2023 to cover new developments related to generative AI following the release of ChatGPT. The European Union's Digital Services Act, adopted in 2022, also includes requirements for algorithmic transparency. These acts made the European Union the first major jurisdiction to implement broad regulations on AI, underscoring an underlying European desire to lead the way in setting the standards and norms. However, the European Union implemented these AI regulations just as European leaders in recent years — including European Commission President Ursula von der Leyen and EU lawmakers from France, Germany and Italy — have pushed back against European regulations more broadly, such as climate and ESG regulations, which they fear may impede Europe's ability to compete economically with major powers like the United States and China. This attitude about regulations was particularly on display with the release of the former European Central Bank President and former Italian Prime Minister Mario Draghi's report on the future of EU competitiveness in September 2024, which laid out a new industrial strategy for Europe and noted, among other things, Europe's challenge in staying economically competitive and the need for a more innovation-friendly regulatory landscape. Amid this broader pushback against potentially burdensome or stifling regulations, the EU AI Act has also faced heat for its potential negative impact on Europe's ability to compete. During deliberations on its latest additions, France, Germany and Italy fought for reduced regulations and sought to exempt foundational models altogether over fears of hindering economic advancement. French President Emmanuel Macron has been particularly vocal in calling for Europe to remove barriers and promote more AI investment and innovation, especially as France houses a number of AI startups, like Mistral AI, that seek to compete with U.S. companies. 

• The updates to the EU AI Act largely center on the categorization of different AI models based on their level of risk, ranging from "unacceptable" and "high risk" to "specific transparency risk" and "minimal risk." The act also considers "systemic risks" of general-purpose AI models, which the bloc says could arise because they are widely used or highly capable. The European Union uses the technical threshold of more than 10 to the power of 25 floating point operations per second in total computing power to train general purpose AI modes as those considered to pose systemic risks. Providers with systemic risks are required to implement risk assessments and mitigations, report serious incidents, ensure cybersecurity, and conduct certain tests and model evaluations. 
• In a statement published Jan. 18, the European People's Party, the largest bloc in the European Parliament that represents center-right parties across the European Union and to which President von der Leyen belongs, called for a two-year pause on the Corporate Sustainability Reporting Directive, Corporate Sustainability Due Diligence Directive, EU taxonomy and Carbon Border Adjustment Mechanism. The party said these requirements are "proving to be excessive and burdensome, with trickle-down effects for European SMEs," underscoring sentiments within Europe to roll back regulations due to alleged negative economic impacts. 
• On Feb. 9 ahead of the AI Action Summit, Macron announced plans for 109 billion euros (about $115 billion) worth of AI investments "in the coming years," saying that "Europe and France must accelerate their investments" in light of competition from China and the United States, indicating concern over falling behind in the AI race. Macron also closed out the first day of the conference discussing the wide availability of nuclear and clean energy in France, which he said would make it ideal for AI infrastructure. 
• In June 2023, following the release of an updated draft of the EU AI Act, C-suite executives from major European companies including Siemens, Carrefour, Renault and Airbus sent an open letter to EU lawmakers raising "serious concerns" about the legislation, saying it "would jeopardize Europe's competitiveness and technological sovereignty." Chief AI scientist at Meta Yann LeCun and founder of British chip company ARM Hermann Hauser also signed the letter. 

As calls to enhance competitiveness continue, European countries will increasingly seek to balance innovation with regulation, opening more opportunities to deploy AI technologies across industry, though not without backlash from those concerned about safety. At the AI Action Summit, European leaders who have advocated for reduced regulations attempted to rally support for their causes, with Macron and EU tech commissioner Henna Virkkunen saying they aimed to cut back on European AI regulation in favor of innovation. European leaders still promise, however, to have the bloc lead AI regulation despite any potential changes to the EU AI Act or its enforcement. To this end, President von der Leyen in a speech at the summit acknowledged the need for European competition but still emphasized that AI "also needs collaboration. And it also needs to be safe." In balancing these values with the need for innovation, the bloc will likely seek to relax regulations in less sensitive areas, like looking at reducing automatic reviews for highly advanced models or ways to streamline compliance mechanisms, in line with Virkkunen's desires to "cut red tape and administrative burdens" to simplify EU rules and ensure they are implemented in a business-friendly way. As more advanced models that meet the computing power threshold for systemic risks are released, the European Union may adjust the threshold to streamline the use and development of these models in the bloc. While doing so may allow European AI companies to develop systems without such regulatory slowdowns, the European Union would still face a long road before it can compete with the United States and China in terms of AI development, as U.S. and Chinese companies still far outpace those in Europe. Weakening regulations in Europe would, however, have much more of an impact in terms of AI deployment and the ability of European companies to take full advantage of AI applications, including from leading models developed in the United States. This would help various sectors, like finance or manufacturing, leverage AI for increased efficiency and scale up their business models to compete economically with other global powers. However, there will be challenges to rolling back aspects of the EU AI Act, with leaders who have been staunch supporters — such as Executive Vice President of the European Commission Margrethe Vestager — likely to push back against any changes. Debating changes will be a long process, and those in favor of AI regulation will continue to push for rules to address their concerns, including the potential for AI to exacerbate inequality and bias, result in job loss, aid threat actors, and harm the environment. 

• The European Union on Feb. 11 announced its own investment of 20 billion euros, or around $20.61 billion, to build large AI data centers. The announcement came alongside European corporations, which also pledged to invest billions in European AI. Indicating corporate views in Europe, the CEO of French AI company Mistral, Arthur Mensch, delivered a speech at the AI Action Summit in which he sought to inspire more European innovation and adoption, saying, "We need Europeans, and we need the world, to actually adopt the technology faster."