Deregulating artificial intelligence will not boost EU tech markets

The European Union is set to take a lighter stance on the regulation of artificial intelligence, in line with a deregulatory proposal issued by the European Commission in November 2025. The plan, already broadly approved by EU countries but awaiting finalisation, would take the EU closer to the approach generally favoured by the United States. Unsurprisingly, Big Tech is supportive. EU policymakers hope the plan will help close the performance gap between the EU and US tech markets.

The plan entails a weakening of tech users’ rights. It makes it easier for AI companies to use sensitive data to train their algorithms. Europeans could be more exposed to discrimination, for example, because an algorithm may use information on their sexual orientation or religious beliefs against their interests. It removes some transparency requirements: if a developer self-assesses that an AI system is not high-risk, she will no longer be obliged to register her application in an EU public database. It broadens the circumstances under which fully automated decisions are legal, even if they result in workers being dismissed by machines rather than humans.

However, there is no convincing evidence that EU safeguarding of fundamental rights is behind the lukewarm performance of European AI markets. The current state of EU tech says more about past industrial choices than about the impacts of digital regulations. For example, in 2003, Europe’s share of global high-tech R&D expenditure was 22%, compared to 55% in the US. Ten years later, it had dropped to 18% in the EU and to 53% in the US. Rather than high-tech, Europe focused on mid-tech sectors, such as car manufacturing. That has nothing to do with European data privacy laws, which have been in effect since 2018, or the AI Act (Regulation (EU) 2024/1689), which is intended to tackle the risks of AI, and which has yet to fully come into force.

Another sign that regulation is not so crucial in determining tech performance comes from China – often seen as the only credible rival to the US in AI. It is estimated that China’s top foundational AI models are now just two months behind the US. However, Chinese tech regulation is far from light touch. Chinese companies must follow a growing set of complex rules.

Companies are required to train chatbots exclusively with government-approved information. China enforces very strict content-moderation laws and significantly limits cross-border data flows. Instead of aiming for stability and setting rules that rarely change, the Chinese approach involves ongoing government decrees to address emerging issues, such as deepfakes. This creates uncertainty, and uncertainty discourages investment.

China shows that even with an allegedly unfavourable regulatory environment, markets can thrive. Other factors, such as energy costs (considered a significant negative by about half of EU businesses) or access to finance, can be more significant. In 2024, the EU’s industrial electricity prices were more than double China’s. From 2013 to 2024, private AI investment totalled $471 billion in the US, followed by China at $119 billion and EU countries at around $50 billion. 

So, is reducing regulatory protection in the EU a price worth paying? Given the marginal role of regulation in explaining tech performance, it is unclear what the European economy would get in return. Privacy protection, for example, may steer companies to change the focus of their innovation efforts (for instance, towards more privacy-friendly applications), but is unlikely to have strong aggregate productivity effects near those implied by access to essential resources, infrastructure and finance.

If anything, reducing protections may undermine confidence in the European digital economy and depress demand for tech services, though demand for AI is likely to grow regardless of regulatory action. EU policymakers should therefore ensure that regulations protect users from harm, rather than overfocusing on the relatively feeble links between regulation and innovation.

Of course, the EU regulatory framework has room for improvement. Regulation may distort competition and favour market concentration, for example, by imposing disproportionate burdens on small companies. Regulation may also fall behind markets, with new risks not anticipated by the lawmakers.

Most AI Act requirements are determined by the risk an application is expected to pose, given its intended purpose. This leaves a gap for unexpected uses that prove harmful in practice (such as ‘nudification’ algorithms or the still unclear, potentially harmful effects of ‘agentic AI’). Public authorities may also struggle to enforce the law and might require new, more effective AI monitoring tools. But though the EU AI regulatory framework needs refinement, it should not be weakened by the misguided belief that looser regulation would benefit the European economy.

This piece is republished in Makronom and Política Exterior.
.