The European Union asserts that its existing AI Act provides sufficient oversight to manage risks from autonomous AI systems. Executive Vice-President Henna Virkkunen stated on October 9, 2026, that the regulation covers the entire life cycle of AI models and allows for continuous monitoring by a panel of 60 experts.
The framework defines systemic risks, including loss of control and large-scale manipulation. Models trained with over 10^25 FLOPs are automatically classified as posing systemic risks. The European Commission’s AI Office holds enforcement powers, including the authority to impose fines of up to €35 million or 7% of a company's global turnover.
The AI Office has already initiated oversight, having sent information requests to more than 30 AI providers as of August 2026. While certain bans have been in effect since February 2025 and transparency rules since August 2026, other requirements have been delayed. Compliance deadlines for high-risk systems are now set for December 2027 and August 2028.
