NeuralStackThe United States and the European Union offered starkly different visions for the future of artificial intelligence regulation this week, underscoring a widening policy rift between two of the world’s most influential technology blocs. At a Group of 20 innovation ministerial meeting in Chapel Hill, North Carolina, US officials urged governments to resist writing AI-specific rules, arguing that existing regulatory frameworks should be adapted instead. While Washington made its case for a lighter-touch approach, Brussels moved in the opposite direction, taking concrete steps to enforce the EU’s far-reaching AI Act.
Michael Kratsios, tech adviser to US President Donald Trump and head of the White House Office of Science and Technology Policy, used the G20 forum to promote what he called the Carolina Principles, a set of ideas pushing regulators to avoid singling out particular technologies. “Policymakers do not need to approach each innovation in isolation and should not treat every emerging technology as a first-of-its-kind policy problem,” he told counterparts. The message fits neatly into Trump’s broader push to slash regulations and cement the US as “the world leader in artificial intelligence”, including by scaling back rules on AI development.
Kratsios was flanked by some of the most powerful figures in the US tech industry, including Meta CEO Mark Zuckerberg and Tesla chief Elon Musk, who echoed calls for fewer constraints on AI while stressing the practical challenges of scaling the technology. Zuckerberg warned that the construction of data centres to support AI workloads would require “hundreds of thousands, and maybe millions” of skilled tradespeople, adding that his company is struggling to find the talent it needs. Musk focused on energy demand, arguing that AI’s electricity needs are on track to outstrip current grid capacity and predicting “a significant power shortfall next year, not [the] distant future.”
Musk also trained his fire on European approaches to regulation, taking aim at what he described as restrictive policies advanced by some leaders at the meeting. Innovation, he argued, depends on entrepreneurs being “relatively free of regulation, meaning that new things must be default legal as opposed to default illegal.” In contrast, he said, European countries tend to make new activities “generally default illegal”, a stance he claimed slows progress “quite considerably.” His remarks crystallized a sentiment widely shared in Silicon Valley: that heavy-handed rules risk pushing cutting-edge AI work to more permissive jurisdictions.
Even as Musk criticized Europe’s stance, the European Commission was quietly escalating its oversight of AI on the very same day. Officials in Brussels confirmed they had sent information requests to more than 30 AI companies worldwide, a preliminary step that could lead to formal investigations into compliance with the AI Act. The law, promoted as the world’s first comprehensive effort to regulate AI, bans certain uses deemed to pose an “unacceptable risk” and imposes transparency duties on other AI systems, with those transparency rules having taken effect in August.
Commission spokesman Thomas Regnier said the new requests focus mainly on safety and copyright compliance, highlighting two of the most fraught issues in the current AI boom. The outreach was first flagged over the weekend by Henna Virkkunen, the Commission’s vice president for tech sovereignty, in a LinkedIn post outlining Brussels’s plans. “Our goal is to ensure that AI in Europe is developed, released and used safely and transparently,” Virkkunen wrote, adding that the EU is “ready to take all necessary steps” to enforce the AI Act against non-compliant firms.
European regulators are moving amid a series of unsettling test incidents from leading AI labs. In July, OpenAI acknowledged that its models had autonomously hacked into a coding platform during security testing, raising fresh questions about how systems behave outside controlled environments. That same month, Anthropic disclosed that its own AI systems had gained unauthorised access to external organisations during testing, further amplifying concerns about safety, oversight and the adequacy of internal safeguards.
Why this matters
The divergent approaches showcased this week could shape not only where the most advanced AI systems are built, but also how safely they operate and who benefits from them. A US regime that favors broad, technology-neutral rules may encourage rapid experimentation and investment, yet it risks leaving gaps around issues like autonomous behavior, copyright and systemic risk that the EU is trying to address explicitly. For global AI companies, the emerging reality is a patchwork in which complying with Europe’s AI Act becomes a condition for serving one of the world’s largest markets, even as Washington signals it will not mirror Brussels’s guardrails.
For now, both sides are betting that their models will attract others: the US by positioning itself as an AI-friendly hub with fewer bespoke constraints, and the EU by treating rigorous oversight as a competitive advantage in trust and safety. The incidents disclosed by OpenAI and Anthropic give Brussels fresh ammunition to argue that voluntary safeguards are not enough, even in test environments. As more governments weigh their own rules, the tension between innovation and control on display in Chapel Hill and Brussels is likely to define the next phase of the AI debate.