Meta and Google executives attack EU AI rules at Stockholm conference

Executives from Meta and Google used a technology conference in Stockholm this week to deliver a coordinated critique of European artificial intelligence regulation, arguing that the bloc's rules are delaying product launches and leaving European consumers with inferior services compared to the United States.

The Stockholm intervention

Speaking at Techarena on Thursday, Chris Yiu, Meta's director of public policy, told an audience of founders and investors that a broad consensus now exists on the problems with European technology regulation. He cited the General Data Protection Regulation as fragmented and the AI Act as overreaching. The net result, he said, is that products arrive late or in watered-down form and European citizens suffer.

Yiu illustrated the point with a pair of Ray-Ban Meta glasses, which use AI to translate speech and describe images for visually impaired users. He described them as a profound, human application of the technology that has been slow to reach Europe because of regulatory complexity. Meta only began rolling out AI features for the glasses in some European countries in November 2024, a delay the company attributes to compliance with what it calls the EU's complex regulatory system.

GDPR and the training data dispute

The glasses are not Meta's first European regulatory headache. The company has previously said that GDPR held up the launch because of questions around its use of Instagram and Facebook user data to train AI models. Irish regulators, who act as Meta's lead supervisory authority under GDPR's one-stop-shop mechanism, have scrutinised the legal basis for processing public posts for model training. The issue remains unresolved for some products, creating a de facto launch barrier that does not exist in the United States.

An Act written before the generative boom

Dorothy Chou, head of public policy at Google DeepMind, offered a different but complementary critique. She noted that the AI Act was first proposed by the European Commission in April 2021, a full 18 months before OpenAI released ChatGPT in November 2022. The legislation was designed for a different technological era, she argued, and regulating on a timescale that does not match the technology creates friction for responsible deployment.

Chou contrasted the EU's approach with the US Inflation Reduction Act, which she said created a better investment environment through subsidies for electric vehicles and other green technologies. The goal, she suggested, should be regulation that ensures responsible application while allowing the industry to thrive.

The general-purpose AI code of practice

The immediate flashpoint is the EU AI Office's code of practice for general-purpose AI models, the systems that underlie products such as OpenAI's GPT family and Google's Gemini. A second draft was published in December 2024. Kent Walker, Google's president of global affairs, told Politico last month that the code was a step in the wrong direction. Joel Kaplan, Meta's newly appointed chief global affairs officer, went further in a live-streamed interview in Brussels earlier this month, saying the company would not sign up in its current form. The rules, he said, go beyond the requirements of the AI Act itself and impose unworkable and technically unfeasible requirements.

Political tailwinds from Washington

The industry's rhetoric has been emboldened by the new Trump administration. At the AI Action Summit in Paris last week, US Vice President JD Vance criticised Europe for focusing on regulation rather than embracing AI's growth potential. The alignment between Big Tech and the White House on this issue is notable: both frame European caution as a strategic error that cedes leadership to the United States and, implicitly, to China.

Venture capital adds its voice

The criticism was not limited to platform executives. Antoine Moyroud, a partner at Lightspeed Venture Partners, which backs the French AI unicorn Mistral, said Europe's narrative around AI is dramatic where the United States is hopeful. He pointed to the $500 billion Stargate investment project announced by the US administration as an example of the kind of positive signalling Europe lacks. The region, he argued, needs to move beyond GDPR and the AI Act and produce technological success stories that excite people about the technology's promise.

The 28th regime and EU Inc

A parallel conversation at Techarena focused on structural barriers to scaling European startups. Luke Pappas, a London-based partner at NEA, said the process of granting equity across borders in Europe is not very easy, making it harder to hire talent across the 27 member states. Standardising equity would dramatically help, he added.

This problem has given rise to the so-called 28th regime proposal, an alternative legal framework that would sit alongside national company laws rather than replace them. The European Company Statute already exists under this model for public limited liability companies. Now a group of founders including Stripe chief executive Patrick Collison and Wise co-founder Taavet Hinrikus are pushing for a new entity type called EU Inc, which would simplify cross-border hiring, fundraising and employee share schemes.