Vance attacks AI regulation in Paris as US, EU and China clash over control

The three-way contest for dominance in artificial intelligence moved into the open on Tuesday at the Grand Palais in Paris, where US Vice President JD Vance delivered a blunt rejection of the European regulatory model and a veiled warning against Chinese partnerships. Speaking to a room of government leaders, tech executives and policymakers, Vance argued that heavy-handed rules would strangle a sector still in its infancy, while framing the American approach as the only guarantee of ideological neutrality and free expression.

Vance frames regulation as an existential threat

"Excessive regulation of the AI sector could kill a transformative sector just as it's taking off," Vance told the summit. The language was deliberate: not merely caution but a prediction of collapse. He positioned the Trump administration as the defender of a permissionless innovation culture, promising that AI systems developed in America would be "free from ideological bias" and that the United States would "never restrict our citizens' right to free speech". The remarks were clearly aimed at the European Union's AI Act, which entered into force last August and imposes tiered obligations based on risk, and at the broader European instinct to regulate first and innovate second.

Vance went further, turning the regulatory argument into a geopolitical one. Without naming China directly, he warned global leaders against striking AI deals with "authoritarian regimes". "Partnering with them means chaining your nation to an authoritarian master that seeks to infiltrate, dig in and seize your information infrastructure," he said. The formulation was striking: it treated data sovereignty and model provenance as security issues, not merely commercial ones, and it signalled that the White House intends to make AI alignment a condition of alliance cooperation.

Europe answers with a 200 billion euro commitment

If the US vice president came to Paris to dismantle the regulatory consensus, European Commission President Ursula von der Leyen arrived to demonstrate financial muscle. She announced an additional 50 billion euros in public funding for the bloc's AI ambitions, drawn from the EU budget and member-state contributions. That sits on top of the 150 billion euros already committed by the European AI Champions Initiative, a coalition of providers, investors and industry groups assembled under Commission auspices. Together the two pots amount to 200 billion euros, a figure von der Leyen presented as proof that Europe can compete on scale without abandoning its regulatory principles.

The European pitch rests on a different premise: that trustworthy AI is a competitive advantage, not a drag. The AI Act's risk-based framework, its proponents argue, will create a single market for compliant systems and give European firms a regulatory passport that non-compliant rivals lack. Critics, including several national governments and parts of the European tech sector, counter that the compliance burden will fall disproportionately on start-ups and that the capital announced on Tuesday is largely a relabelling of existing Horizon Europe and national programmes rather than fresh money.

China pushes open access while restricting domestic data

Chinese Vice Premier Zhang Guoqing attended the summit in person, a signal of how seriously Beijing treats the standard-setting conversation. According to the state news agency Xinhua, Zhang said China is "willing to work with other countries to safeguard security and share achievements in the field, in order to build a community with a shared future for mankind". The language is familiar from Chinese diplomacy on climate, trade and digital infrastructure: an offer of cooperation that positions China as a generous stakeholder rather than a strategic rival.

The reality is more complicated. Beijing promotes open-source AI models, most visibly through the rapid rise of DeepSeek, whose chatbot has rattled Silicon Valley and prompted calls in the US Congress for restrictions on its use, while maintaining strict controls over domestic data flows and model deployment. Chinese officials on Monday condemned Western efforts to restrict access to AI tools, framing such moves as containment disguised as security. The tension between China's open-source outreach and its closed data environment is likely to become a central fault line in any future multilateral governance talks.

NATO warns of a control problem that regulation cannot yet solve

Admiral Pierre Vandier, who oversees modernisation for NATO, brought a military perspective that cut across the commercial and diplomatic rhetoric. "I think one day we will have to find ways to control AI or else we will lose control of everything," he said. The remark underscored a concern that has been growing inside the alliance for years: that the integration of AI into targeting, logistics, surveillance and decision-support systems is outpacing the ability of governments to verify, validate and, if necessary, interrupt those systems.

NATO's 2024 revised AI strategy identifies six principles, lawfulness, responsibility, accountability, reliability, governability and traceability, but stops short of proposing binding verification regimes. Member states remain divided on whether such regimes should be negotiated within the alliance, at the United Nations, or through a new dedicated body. Vandier's comment suggests the military leadership believes the current voluntary framework is temporary and that the summit's focus on commercial regulation misses the harder problem of autonomous weapons and escalation risks.

The private sector battle overshadows the public diplomacy

While ministers and commissioners debated frameworks, a parallel drama unfolded in the corridors. A consortium led by Elon Musk, who now heads the US Department of Government Efficiency, tabled a 97.4 billion dollar bid for the nonprofit entity that controls OpenAI. Sam Altman, OpenAI's chief executive and a attendee at the summit, rejected the offer outright. The bid is widely seen as an extension of Musk's long-running dispute with Altman over the direction of the company he co-founded, but its timing, during a summit nominally about global governance, illustrated how concentrated power over frontier models has become.

OpenAI's unusual structure, a capped-profit subsidiary governed by a nonprofit board, was designed to align the company with a broad public benefit. Musk's attempt to acquire the governing entity would effectively privatise that governance. Altman's refusal, and the fact that both men were in Paris for the same event, turned the summit into a stage for a corporate control struggle that has implications for how the most capable models are stewarded, priced and restricted.

No joint declaration, no shared red lines

The summit produced no communiqué, no agreed principles and no commitment to a follow-up negotiation. That absence is itself a finding. The last major AI gathering, at Bletchley Park in November 2023, yielded a declaration signed by 28 countries and the EU acknowledging shared risks. The Seoul summit in May 2024 added a pledge to develop safety institutes. Paris, by contrast, exposed the fractures: the US rejects the premise of pre-market regulation, the EU treats regulation as a market-shaping tool, and China offers open-source code while locking down data. The only convergence was on the need for more investment, and even there the sources and conditions differ radically.