Bloc-level rules governing artificial intelligence were described by EU officials as adequate for managing the hazards tied to next-generation AI systems, among them the prospect that highly capable models might act in ways their designers neither intended nor foresaw. Virkkunen, who holds the European Commission's executive vice-presidency covering technological sovereignty, security and democracy, expressed confidence that the bloc had adequate tools and capacity to handle those threats.

Broader unease across the international community over episodes tied to both OpenAI and Anthropic formed the backdrop against which her remarks arrived.

Under the EU's AI Act, a structured set of obligations covers risk evaluation, ongoing oversight of powerful models, and corporate accountability as artificial intelligence technology continues to evolve. A framework built on graduated risk tiers governs how the EU's AI Act — ratified in 2024 — applies its requirements across different categories of artificial intelligence applications. Proponents describe it as among the most far-reaching regulatory frameworks for artificial intelligence yet produced anywhere in the world, though assessments of its scope vary.

Virkkunen gave no ground to the argument, advanced by some technology companies, that events had already rendered the regulations obsolete. Her position held that legislators had built into the framework a capacity for reassessing emerging threats, drawing on outside expertise, and maintaining oversight of AI systems beyond the point of their first formal review. Oversight bodies would set out the criteria by which companies are required to evaluate their systems, along with the timelines governing how those assessments are conducted.

Questionnaires went out from the European Commission to upwards of 30 artificial intelligence firms toward the end of August, pressing each one to account for the protective and security protocols built into their systems. No names of the firms in question have been released by the Commission. Questions about transparency and copyright obligations arose during the follow-up portion of the session.

Virkkunen added that AI developers based in China had received similar requests, pointing out that both the United States and China are home to some of the most capable models on the market today. Formal investigations may follow from those submissions. Breaches of the applicable rules leave companies open to financial penalties reaching as high as 7% of their total worldwide annual turnover.

A review of the submitted responses is now under way at the Commission, Virkkunen indicated.

Questions about how far governments can go in overseeing advanced AI technology — without placing undue constraints on the industry's capacity to develop — form the backdrop against which the EU's stance has taken shape. Questions about how to govern, verify, and maintain human control over AI systems have drawn growing attention alongside the technology's advancing capabilities. European regulators face a dual challenge: adapting oversight tools to threats that are still taking shape while ensuring the rules already on the books retain their practical effect.

What the Commission's review process ultimately yields will indicate whether its existing framework holds up or whether additional measures are warranted.

Plans for a digital levy targeting large technology companies drew a response from Virkkunen, who set out the administration's position on the matter. The final shape of those revenue streams will rest with EU governments, whose negotiations over the bloc's next long-term budget — covering 2028 to 2034 — are expected to address the matter. Questions about how Europe weighs artificial intelligence oversight against technology policy and funding allocations have come into sharper focus through this debate, reflecting the continent's broader push to shape its own digital trajectory.