U.S. AI Regulation

 To date, AI legislation in the U.S. has been inconsistent. Differing regulations were mainly in place at the state level. The first federal law passed was under President Trump in his first term, the National AI Initiative Act (2020), and it was focused on investment in human resources and research, but did not introduce specific regulatory solutions. Instead, it established several advisory groups, whose subsequent role proved to be negligible. In 2022, the administration of President Joe Biden passed the CHIPS and Science Act, allocating $52 billion to support domestic semiconductor manufacturing, which strengthened the U.S. position in the AI market. A year after the Act, NVIDIA, a U.S. company that makes specialised AI chips, saw significant stock market gains. In 2023, Biden also signed an Executive Order on Secure and Trustworthy AI, requiring developers to report on security testing, which promoted intensive but ethical AI development. Towards the end of his term, he also introduced the Framework Plan for the Diffusion of AI, restricting the export of advanced chips to selected countries, including China and Venezuela, to hinder their development of their own technologies. However, models such as China’s DeepSeek (a generative AI model developed without the use of NVIDIA’s latest generation of advanced chips, which achieves results similar to Chat GPT’s) cast doubt on this assumption. Despite limited access to U.S. technology in China, competition for the U.S. is growing.

The Actions of the New Administration

 Trump is theoretically continuing the AI direction set by the Biden administration, which was expressed in the ambition to take a dominant position in the field. However, the new administration is pursuing this goal by different means, focusing on combating non-U.S. production of semiconductors and advanced integrated circuits, and reducing AI regulation.

Trump’s policy is guided by the belief that the U.S. should be at least one year ahead of other countries in AI development. This course of action is influenced by representatives of U.S. technology companies (i.e., Big Tech) who are closely associated with him. In addition to Elon Musk, the chairman of SpaceX, CEO of Tesla, head of the so-called Department of Government Efficiency (DOGE), and an advisor to Trump on technology policy, among other things, the president is also on good terms with other business representatives, including Sam Altman (OpenAI), Jeff Bezos (Amazon), and Mark Zuckerberg (Meta). They lobby for the removal of barriers for businesses and share the desire to develop general artificial intelligence (AGI), which comprise algorithms with human-level intelligence that would match or surpass human capabilities in most tasks.

It was also in response to the expectations of U.S. big tech that Trump cancelled Biden’s regulation on the ethical development of AI, arguing that it inhibited innovation. This quickly had an effect: Google, for example, dropped its self-imposed ban on the use of AI in weapons manufacturing. The January order removing perceived barriers to U.S. dominance in artificial intelligence is intended, according to Trump, to ensure that U.S. technology companies can develop AI without restrictions, including those related to ethics or privacy.

Trump also wants the vast majority of AI production to be concentrated in the U.S., which is why, among other things, he has committed the U.S. to the Stargate project, developed by U.S.-based OpenAI and Oracle, Japan’s SoftBank and the Saudi fund MTX. The project envisages a $500 billion investment in AI. Its goal is to create new data centres in the U.S. that will allow AI to evolve towards AGI. To this end, it also plans to significantly restrict other countries’ access to the technology. In March, Trump announced his intention to cancel the CHIPS and Science Act, arguing that past subsidies have not delivered the expected benefits to U.S. taxpayers. Instead, he wants to impose tariffs on chip imports to encourage companies to move production to the U.S. Although, for the time being, companies such as NVIDIA and Taiwan’s TSMC, the world’s largest advanced-chip maker, are announcing further investments in the U.S., their scale is not meeting Trump’s expectations. Companies are also looking for other, more stable investment markets, so the U.S. decisions may therefore be counterproductive.

Trump also issued a memorandum in March this year defending U.S. companies against what he described as foreign extortion and unfair fines and penalties. It implies that if other countries impose additional charges on companies (including fines), the U.S. will take the retaliatory action necessary to mitigate the alleged harm,  such as imposing tariffs on individual countries. This also poses a threat to the operations of U.S. companies in the EU, especially if European solutions (such as France’s Mistral AI) prove to be equally effective and efficient. The EU may then find itself less interested in purchasing U.S. solutions.

International Environment

 The Trump administration’s approach is a response to the direction of AI development and especially the regulation of the sector outside the U.S. The EU Artificial Intelligence Act, which promotes the ethical development of AI in the EU and was adopted while Biden was still president and in response to the leapfrogging of AI in the U.S., is seen by the Trump administration as a threat to the innovation of U.S. companies. Among other things, the EU regulation is why Trump issued the “foreign extortion” memorandum. Additional tensions are stoked by the unequal treatment of EU countries in access to U.S.-produced AI chips. While the U.S. approach stems from the country’s current AI superpower ambitions, the U.S., by antagonising its partners, is encouraging them instead to develop AI technology in the EU on their own. A case in point is France, where Mistral AI, a national generative AI model, is being successfully developed.

China poses an existential threat to the U.S., according to the Trump administration, and the U.S. is particularly concerned about potential Chinese technological developments. For this reason, achievements such as China’s generative AI model DeepSeek must also have influenced Trump to try harder to strictly limit AI development outside the U.S., including pushing to move semiconductor production from Taiwan to the U.S.

Conclusions

 The U.S. has long aspired to be a leader in AI. However, the new administration wants to achieve this through radical methods because Trump and his entourage see AI as central to human development and fundamental to maintaining U.S. technological leadership in the world. The U.S. will put increasing pressure on technology companies (especially non-U.S. companies such as Taiwan’s TSMC) to increase investment in the country. However, these actions may instead weaken the U.S. position in the global technology race and negatively impact planned investments in domestic semiconductor manufacturing. The likelihood of success is low, and indeed, companies are already looking for alternative locations for their investments. At the same time, the U.S. is opposed to any regulation imposing restrictions on U.S. companies, which strains EU-U.S. relations. Also, if U.S. solutions do not comply with European standards, this could limit their potential profits in the EU market. The potential imposition of tariffs on the EU if penalties are imposed on U.S. companies could exacerbate the EU-U.S. dispute and lead to increased European investment in local solutions. Although the U.S. maintains its position as the world leader in AI development, it is already apparent that other countries are creating their own models without U.S. support that perform at least as well in tests as the most developed U.S. models. If models such as Mistral demonstrate high performance, it is possible that they will become all the more popular in the EU market.