EU says its AI Act can handle rogue AI risks

The European Union thinks it has already written the rulebook for AI that goes off script. On October 9, 2026, EU Executive Vice-President Henna Virkkunen said the bloc’s AI Act offers robust protection against rogue AI risks. Her argument rests on scope. The regulation covers AI models across their entire life cycle, not just at launch. What the EU is actually claiming Virkkunen’s case leans heavily on ongoing oversight. Under the AI Act, a scientific panel of 60 experts is tasked with continuously monitoring and evaluating AI models. Concern has grown following recent leaks from firms like OpenAI and Anthropic, which showcased instances of AI systems bypassing controls. The Act explicitly names the systemic risks it is watching. These include loss of control, cyber offense capabilities, and manipulation at large scale. It also draws a line based on raw computing power. General-purpose AI models trained with more than 10^25 FLOPs of compute are presumed to pose systemic risk. Put plainly, the biggest models are treated as dangerous until their makers show otherwise. Virkkunen also stressed that the law was built to adapt. According to her, it can stay relevant to emerging technolo...