U.S. could feel effects of EU AI Act as companies comply
The U.S. may be making a deregulatory push on AI, but the EU AI Act means large U.S. AI developers must comply with AI regulations that will affect their models regardless.
Though America's AI Action Plan moves away from regulations viewed as burdensome to AI innovation, the European Union isn't slowing down enforcement of its AI Act. As a result, the U.S. could still feel the effects of AI regulation.
Multiple AI model developers, including OpenAI, Microsoft, Cohere, Amazon, Google, Anthropic and Mistral AI, signed the recently released General-Purpose AI Code of Practice (GPAI). By signing the GPAI, AI model developers are demonstrating their compliance with the EU AI Act. The GPAI aims to reduce companies' administrative burdens and provide more legal certainty for compliance.
"The AI Act puts a lot of constraints or requirements on companies," said Mia Hoffman, a research fellow at Georgetown University's Center for Security and Emerging Technology. "For instance, to implement model evaluations or to do a risk assessment for general-purpose AI. But there is very little detail as to what that actually means in practice." Hoffman said the GPAI helps model providers know what is expected of them under the EU AI Act.
As companies comply with the EU AI Act to preserve business operations in the European market, Hoffman said the protections offered through the EU regulations will also apply to consumers in the U.S. because it's likely the same models will be released in both markets.
As much as they might try to approach this deregulatory agenda in the U.S., it does not prevent U.S. AI companies from having to comply with the European Union's rules.
Mia HoffmanResearch fellow, Center for Security and Emerging Technology, Georgetown University
AI developers could create different versions of their AI systems for different markets. However, Hoffman said this option is unlikely due to the high costs associated with training the largest and most advanced models, like OpenAI's GPT-5 and xAI's Grok 4.
"As much as they might try to approach this deregulatory agenda in the U.S., it does not prevent U.S. AI companies from having to comply with the European Union's rules," she said.
Federal deregulatory push not slowing other AI laws
Meta is a notable holdout from signing the GPAI. Joel Kaplan, Meta's chief global affairs officer, said in a statement that the GPAI introduces legal uncertainties for businesses and measures that go beyond the scope of the EU AI Act.
The GPAI is only one approach to complying with the EU AI Act, said Ayesha Bhatti, head of digital policy for the U.K. and EU at the Information Technology and Innovation Foundation's Center for Data Innovation. Elon Musk's xAI, for example, signed the Safety and Security Chapter within the GPAI but must demonstrate its compliance with the EU AI Act's transparency and copyright requirements through alternative means.
"The fact that Meta and others haven't signed up to the code means it doesn't work for everyone across the board," Bhatti said. "That doesn't mean they aren't going to comply with the AI Act, it just means they'll be taking a different approach to comply."
The EU AI Act could have unintended consequences, such as discouraging companies from exploring new AI-powered business models due to the high cost of compliance that comes with it, Bhatti said.
Georgetown's Hoffman said concerns about EU regulations hindering innovation have arisen, particularly following the September 2024 report by Mario Draghi, former European Central Bank president, highlighting the need for more flexible regulation to improve EU competitiveness. As a result, Hoffman said there has been an attitude shift in the EU toward fostering competitiveness and promoting a European digital ecosystem for AI development.
Still, while the EU seeks to guide businesses through regulations to support growth and innovation, as well as ease compliance burdens, Hoffman said they aren't looking to postpone implementation of laws like the EU AI Act.
"There is still a very strong commitment at the EU level that this regulation is happening, it will be implemented, and everybody who wants to enter the EU market needs to comply with it," she said.
It's not just global AI regulations that businesses must adhere to. Companies must also comply with multiple U.S. state laws targeting AI, which may disrupt the federal deregulatory approach.
OpenAI sent a letter to California Governor Gavin Newsom to align the state's regulations on frontier AI models with the GPAI to make compliance less burdensome. U.S. companies face the growing challenge of complying with different versions of AI laws across U.S. states as they adopt their own AI rules.
Policymakers included a proposal to place a moratorium on state AI laws in the recently passed U.S. spending bill. However, the measure failed to pass.
Makenzie Holland is a senior news writer covering big tech and federal regulation. Prior to joining Informa TechTarget, she was a general assignment reporter for the Wilmington StarNews and a crime and education reporter at the Wabash Plain Dealer.
