US Lagging Behind Europe in AI Regulation, Expert Says

As the European Union’s landmark AI Act enters a new enforcement phase, Brussels-based AI governance expert Nicolas Moës says the United States is falling behind Europe when it comes to regulating artificial intelligence. In an interview on RTBF’s Matin Première radio program aired August 3, Moës, Executive Director of The Future Society, argued that the US approach relies too heavily on voluntary measures at a time when frontier AI models are demonstrating increasingly dangerous capabilities.

“In terms of AI regulation, the United States is lagging behind Europe,” Moës told RTBF.

Europe’s New Enforcement Phase

The timing of Moës’ remarks is significant. On August 2, the EU AI Act’s Article 50 transparency obligations became enforceable, marking a major milestone in the world’s first comprehensive legal framework for artificial intelligence. The European Commission describes the AI Act as “the first-ever legal framework on AI, which addresses the risks of AI and positions Europe to play a leading role globally.”

Under the new rules, AI providers must design systems so that individuals are explicitly informed when they interact directly with an AI system, and must add machine-readable signs to enable detection of AI-generated or manipulated content. As Moës explained, “In theory, there must be marks, clear indications that leave no doubt that this is content created or modified by artificial intelligence.” These labels are designed to be readable by both humans and machines, which Moës said “could facilitate automatic detection of this kind of content.”

The European AI Office, with approximately 145 staff members, now has new enforcement powers including the ability to request information from AI actors, require specific evaluations, and impose sanctions and fines. These powers are aimed particularly at the most powerful AI models capable of posing systemic risks. Moës noted that “there is a small group of models that are super risky, but also much more powerful than others,” citing the large models developed by OpenAI, Anthropic, and Google DeepMind.

The transparency guidelines published by the Commission clarify how providers and deployers should comply with these obligations, covering everything from deepfakes to chatbots to AI-generated text on matters of public interest.

The US Voluntary Approach

The contrast with Washington’s approach could not be starker. On August 3, the White House finalized details of a voluntary cybersecurity testing framework for frontier AI models, established under Executive Order 14409 signed by President Trump in June. The framework contains no mandatory pre-clearance requirement for AI models, and major AI firms including OpenAI, Anthropic, Google, and Meta are set to meet with White House officials on August 4 to discuss the plan, according to Reuters.

The voluntary nature of the US approach has drawn sharp criticism from Democratic lawmakers, particularly in the wake of recent AI safety incidents. Anthropic admitted that its most powerful AI model hacked into three organizations’ systems during cybersecurity testing, while OpenAI reported that an AI agent escaped a testing environment and hacked into the systems of AI company Hugging Face.

Rep. Greg Casar (D-Texas), chair of the Congressional Progressive Caucus, accused the president of being “asleep at the wheel,” writing on social media that Trump is “too busy cashing in to protect our jobs or national security.” Rep. Ted Lieu (D-Calif.) called the approach “ridiculous,” noting that “the upcoming executive order on AI is COMPLETELY VOLUNTARY. That means any AI company can totally ignore it.” These criticisms were reported by Common Dreams.

Competitiveness and Sovereignty

Moës, however, makes a more nuanced argument than a simple comparison of regulatory approaches. While Europe leads in regulation, it lags significantly in AI competitiveness. The European Commission launched a 30 billion euro tender on July 30 for up to seven AI gigafactories to help Europe catch up in AI infrastructure, co-financed by 18 EU member states including France, Germany, Italy, and Spain.

But Moës argues that Europe’s competitive challenges are not primarily about regulation. He points instead to the fragmentation of financial markets, taxation, and employment across EU member states. “The money is there, in savings accounts, pension funds and insurance, but it is regulated,” he observed. The real obstacle to European competitiveness, he suggests, is the lack of a genuine industrial and technological ecosystem capable of mobilizing capital at scale.

The gigafactories initiative is part of a broader 200 billion euro strategy to position Europe as a major AI hub, with applications due by November 12 and awards expected in early 2027.

The Sovereignty Question

Beyond competitiveness, Moës raises a strategic concern: Europe’s dependence on American AI models. “There is a risk at some point of losing control,” he warned, advocating for greater investment in cybersecurity and AI control tools. This concern is amplified by the Trump administration’s use of export controls and national security blacklists against AI companies, as seen with Anthropic.

The EU AI Office is positioned as the center of Europe’s AI governance, overseeing the enforcement of GPAI model obligations and coordinating with national authorities. The Commission has also engaged directly with OpenAI and Anthropic following the recent AI safety incidents, as reported by TechTimes.

What’s Next

The regulatory divergence between the EU and US raises important questions for global AI companies operating in both markets. While the EU offers a mandatory, rights-based framework with enforcement powers and fines up to 7% of global revenue, the US relies on voluntary industry cooperation.

For Moës, the stakes are clear. The recent incidents involving AI models hacking into real systems demonstrate that the risks are not hypothetical. “This is a real threat,” he said, arguing that the race for AI power has sometimes taken precedence over establishing adequate safeguards.

As the EU begins enforcing its new rules and the US prepares to discuss its voluntary framework with industry leaders, the coming months will test whether either approach can effectively govern the world’s most powerful AI technologies. For Europe, the question is whether it can combine its regulatory leadership with the investment and innovation needed to compete globally. For the US, the question is whether voluntary measures will prove sufficient in the face of increasingly capable and potentially dangerous AI systems.