How AI regulation could shake out in 2025

The U.S. political landscape is set to undergo some shifts in 2025 — and those changes will have some major implications for the regulation of artificial intelligence.

President-elect Donald Trump will be inaugurated on Jan. 20. Joining him in the White House will be a raft of top advisors from the world of business — including Elon Musk and Vivek Ramaswamy — who are expected to influence policy thinking around nascent technologies such as AI and cryptocurrencies.

Across the Atlantic, a tale of two jurisdictions has emerged, with the U.K. and European Union diverging in regulatory thinking. While the EU has taken more of a heavy hand with the Silicon Valley giants behind the most powerful AI systems, Britain has adopted a more light-touch approach.

In 2025, the state of AI regulation globally could be in for a major overhaul. CNBC takes a look at some of the key developments to watch — from the evolution of the EU’s landmark AI Act to what a Trump administration could do for the U.S.

Although it’s not an issue that featured very heavily during Trump’s election campaign, artificial intelligence is expected to be one of the key sectors set to benefit from the next U.S. administration.

For one, Trump appointed Musk, CEO of electric car manufacturer Tesla, to co-lead his “Department of Government Efficiency” alongside Ramaswamy, an American biotech entrepreneur who dropped out of the 2024 presidential election race to back Trump.

Matt Calkins, CEO of Appian, told CNBC Trump’s close relationship with Musk could put the U.S. in a good position when it comes to AI, citing the billionaire’s experience as a co-founder of OpenAI and CEO of xAI, his own AI lab, as positive indicators.

“We’ve finally got one person in the U.S. administration who truly knows about AI and has an opinion about it,” Calkins said in an interview last month. Musk was one of Trump’s most prominent endorsers in the business community, even appearing at some of his campaign rallies.

There is currently no confirmation on what Trump has planned in terms of possible presidential directives or executive orders. But Calkins thinks it’s likely Musk will look to suggest guardrails to ensure AI development doesn’t endanger civilization — a risk he’s warned about multiple times in the past.

“He has an unquestioned reluctance to allow AI to cause catastrophic human outcomes – he’s definitely worried about that, he was talking about it long before he had a policy position,” Calkins told CNBC.

Currently, there is no comprehensive federal AI legislation in the U.S. Rather, there’s been a patchwork of regulatory frameworks at the state and local level, with numerous AI bills introduced across 45 states plus Washington D.C., Puerto Rico and the U.S. Virgin Islands.

The European Union has so far been the only jurisdiction globally to push forward with comprehensive statutory rules for the AI industry. Earlier this year, the bloc’s AI Act — a first-of-its-kind AI regulatory framework — officially entered into force.

The law isn’t yet fully in force yet, but it’s already causing tension among large U.S. tech companies, who are concerned that some aspects of the regulation are too strict and may quash innovation.

In December, the EU AI Office, a newly created body overseeing models under the AI Act, published a second-draft code of practice for general-purpose AI (GPAI) models, which refers to systems like OpenAI’s GPT family of large language models, or LLMs.

The second draft included exemptions for providers of certain open-source AI models. Such models are typically available to the public to allow developers to build their own custom versions. It also includes a requirement for developers of “systemic” GPAI models to undergo rigorous risk assessments.

The Computer & Communications Industry Association — whose members include Amazon, Google and Meta — warned it “contains measures going far beyond the Act’s agreed scope, such as far-reaching copyright measures.”

The AI Office wasn’t immediately available for comment when contacted by CNBC.

It’s worth noting the EU AI Act is far from reaching full implementation.

As Shelley McKinley, chief legal officer of popular code repository platform GitHub, told CNBC in November, “the next phase of the work has started, which may mean there’s more ahead of us than there is behind us at this point.”

For example, in February, the first provisions of the Act will become enforceable. These provisions cover “high-risk” AI applications such as remote biometric identification, loan decisioning and educational scoring. A third draft of the code on GPAI models is slated for publication that same month.

European tech leaders are concerned about the risk that punitive EU measures on U.S. tech firms could provoke a reaction from Trump, which might in turn cause the bloc to soften its approach.

Take antitrust regulation, for example. The EU’s been an active player taking action to curb U.S. tech giants’ dominance — but that’s something that could result in a negative response from Trump, according to Swiss VPN firm Proton’s CEO Andy Yen.

“[Trump’s] view is he probably wants to regulate his tech companies himself,” Yen told CNBC in a November interview at the Web Summit tech conference in Lisbon, Portugal. “He doesn’t want Europe to get involved.”

One country to watch for is the U.K. Previously, Britain has shied away from introducing statutory obligations for AI model makers due to the fear that new legislation could be too restrictive.

However, Keir Starmer’s government has said it plans to draw up legislation for AI, although details remain thin for now. The general expectation is that the U.K. will take a more principles-based approach to AI regulation, as opposed to the EU’s risk-based framework.

Last month, the government dropped its first major indicator for where regulation is moving, announcing a consultation on measures to regulate the use of copyrighted content to train AI models. Copyright is a big issue for generative AI and LLMs, in particular.

Most LLMs use public data from the open web to train their AI models. But that often includes examples of artwork and other copyrighted material. Artists and publishers like the New York Times allege that these systems are unfairly scraping their valuable content without consent to generate original output.

To address this issue, the U.K. government is considering making an exception to copyright law for AI model training, while still allowing rights holders to opt out of having their works used for training purposes.

Appian’s Calkins said that the U.K. could end up being a “global leader” on the issue of copyright infringement by AI models, adding that the country isn’t “subject to the same overwhelming lobbying blitz from domestic AI leaders that the U.S. is.”

Lastly, as world governments seek to regulate fast-growing AI systems, there’s a risk geopolitical tensions between the U.S. and China may escalate under Trump.

In his first term as president, Trump enforced a number of hawkish policy measures on China, including a decision to add Huawei to a trade blacklist restricting it from doing business with American tech suppliers. He also launched a bid to ban TikTok,which is owned by Chinese firm ByteDance, in the U.S. — although he’s since softened his position on TikTok.

China is racing to beat the U.S. for dominance in AI. At the same time, the U.S. has taken measures to restrict China’s access to key technologies, mainly chips like those designed by Nvidia, which are required to train more advanced AI models. China has responded by attempting to build its own homegrown chip industry.

Technologists worry that a geopolitical fracturing between the U.S. and China on artificial intelligence could result in other risks, such as the potential for one of the two to develop a form of AI smarter than humans.

Max Tegmark, founder of the nonprofit Future of Life Institute, believes the U.S. and China could in future create a form of AI that can improve itself and design new systems without human supervision, potentially forcing both countries’ governments to individually come up with rules around AI safety.

“My optimistic path forward is the U.S. and China unilaterally impose national safety standards to prevent their own companies from doing harm and building uncontrollable AGI, not to appease the rivals superpowers, but just to protect themselves,” Tegmark told CNBC in a November interview.

Governments are already trying to work together to figure out how to create regulations and frameworks around AI. In 2023, the U.K. hosted a global AI safety summit, which the U.S. and China administrations both attended, to discuss potential guardrails around the technology.

– CNBC’s Arjun Kharpal contributed to this report