The EU spent years writing rules for powerful AI. Now regulators are beginning to ask AI companies to prove they are following them.

WHAT’S HAPPENING

The European Commission has begun formally requesting information from providers of general-purpose AI models as enforcement powers under the EU AI Act take effect.

The requests focus on areas including model security, testing, post-release monitoring and transparency around training content.

Major AI companies including OpenAI, Anthropic and Google were reportedly among the providers contacted.

WHY IT MATTERS

This marks an important transition for European AI regulation.

The AI Act is no longer only a framework describing what companies are supposed to do.

Regulators are beginning to demand evidence.

Providers that fail to comply with formal information requests, or provide incorrect or misleading information, can face enforcement action and financial penalties.

WHO BENEFITS

European regulators gain direct access to information needed to evaluate whether powerful AI systems meet legal requirements.

Users and businesses could benefit from stronger testing, monitoring and accountability around widely used AI models.

AI companies with mature compliance systems may gain an advantage if regulation raises the cost of operating less-transparent models.

WHO LOSES

AI providers with weak documentation, testing or transparency systems face greater regulatory exposure.

Smaller developers may also face higher compliance costs as Europe demands more documentation around increasingly capable models.

WHAT HAPPENS NEXT

More information requests, evaluations and corrective actions are likely as the EU AI Office expands supervision of general-purpose AI.

The key question is no longer whether Europe can write AI rules.

It is whether regulators can enforce those rules against the companies building the world’s most powerful AI systems.