European tech companies and their global counterparts are increasingly delaying or withholding AI model releases in the EU due to strict regulatory requirements, according to new research that quantifies the real-world cost of Europe's approach to AI governance. A comprehensive analysis of 375 large language models (LLMs) released over the past eight years reveals that regulatory friction is creating a meaningful gap between what European users can access and what's available elsewhere.

Why Are AI Companies Delaying Releases in Europe?

The study, released by Governance.AI, examined language models from major companies including Meta, Google, OpenAI, and Anthropic between June 2018 and May 2026. The findings paint a clear picture of regulatory burden: at least 11% of model releases were delayed or not released at all in the EU, compared to 7% in the UK. Of the 68 documented cases of delays and non-releases, regulatory factors were the primary cause in 56 instances.

Data protection regulations emerged as the single biggest obstacle, particularly for non-text AI capabilities like image generation, audio processing, and real-time video analysis. The European Union's General Data Protection Regulation (GDPR), combined with the Digital Markets Act (DMA) and the AI Act, creates a complex compliance landscape that companies say is difficult to navigate. One notable example: Anthropic's Claude 3 Opus web application experienced a 71-day delay in the EU compared to its US release.

Meta faced the steepest regulatory headwinds, with more than a quarter (26%) of its releases delayed or withheld in the EU and 15% in the UK. The disparity between EU and UK outcomes is particularly telling, since both regions share similar data protection laws through the GDPR. The difference appears to stem from the EU's more aggressive enforcement approach and slower clarification of how existing data protection rules apply to AI model training and deployment.

What's Driving the Regulatory Bottleneck?

The EU's regulatory framework for AI has evolved rapidly in recent years. The AI Act, adopted in 2024, introduced new compliance requirements for high-risk AI systems. The DMA, enforced starting in 2023, imposes additional obligations on large digital platforms. Together with the GDPR's strict data handling requirements, these rules create overlapping compliance demands that companies say lack clear guidance on practical implementation.

The rigidity of Europe's data protection framework presents a particular challenge for AI development. Unlike the US approach, which has historically favored lighter-touch regulation, the EU's rules were designed before large language models became mainstream. As a result, companies face uncertainty about how to legally train AI systems on European data while respecting privacy protections.

The EU does recognize this tension. Policymakers are currently considering the Digital Omnibus, a legislative proposal designed to make data rules more workable for AI development. However, this effort faces a countervailing pressure: the EU is simultaneously reviewing copyright protections and the AI Act's copyright provisions to safeguard authors' rights. If applied rigidly, these copyright rules could further restrict AI model availability in Europe.

How Can European Policymakers Balance Innovation and Protection?

• Clarify Data Protection Rules: The EU should issue detailed guidance on how GDPR applies to LLM training and deployment, reducing legal uncertainty that currently delays product launches and forces companies to make conservative compliance choices.
• Align Regulatory Timelines: Coordinate the implementation schedules of the AI Act, DMA, and data protection enforcement to prevent overlapping compliance deadlines that stretch company resources and slow innovation cycles.
• Monitor Regulatory Impact: Establish metrics to track how regulatory barriers affect European access to frontier AI models, ensuring that policymakers can adjust rules if delays become severe enough to harm competitiveness or consumer welfare.

John Lidiard, a UK AI policy researcher at Governance.AI, emphasized the stakes of this regulatory approach. "It's important that policymakers in the EU and the UK are calibrated to the risk of regulatory barriers causing delays for their citizens and businesses in accessing the latest AI models," he stated. "Our report finds that European regulation, primarily the GDPR, led frontier AI companies to sometimes delay model releases or in some cases not release models at all to the EU and UK. Policymakers should consider delays to model access as a factor when implementing and designing AI-related regulations."

"It's important that policymakers in the EU and the UK are calibrated to the risk of regulatory barriers causing delays for their citizens and businesses in accessing the latest AI models. Our report finds that European regulation, primarily the GDPR, led frontier AI companies to sometimes delay model releases or in some cases not release models at all to the EU and UK. Policymakers should consider delays to model access as a factor when implementing and designing AI-related regulations," stated John Lidiard.

John Lidiard, UK AI Policy Researcher at Governance.AI

The broader context for this debate involves Europe's evolving approach to digital sovereignty. Separately, the EU is advancing the Cloud and AI Development Act (CADA), part of a wider Tech Sovereignty Package that includes initiatives like the Chips Act 2.0 and an EU Open-Source Strategy. CADA introduces a harmonized EU-wide framework measuring cloud and AI sovereignty across four strict "Union assurance levels" based on physical location, operational independence, ownership, and supply-chain control. This framework reflects Europe's determination to reduce dependence on non-EU cloud and AI infrastructure providers.

The tension between these two policy directions is significant. While CADA and related sovereignty measures aim to strengthen European AI capacity, the strict regulatory requirements that delay model releases may inadvertently slow the very innovation Europe seeks to foster. The challenge for EU policymakers is calibrating regulations that protect privacy and competition without creating such high barriers that European consumers and businesses fall further behind in accessing cutting-edge AI tools.

The study's findings suggest that the cost of regulatory caution is already measurable. As frontier AI companies continue to release new models at an accelerating pace, the gap between what Europeans can access and what's available globally may widen further unless policymakers find ways to streamline compliance pathways without compromising their core regulatory objectives.