EU says its AI Act can handle rogue AI risks
Crypto Briefing 2026-10-09 23:53:56
Context: The European Union's AI Act, introduced in 2024, is designed to mitigate risks associated with artificial intelligence, including rogue AI threats. On October 9, 2026, EU Executive Vice-President Henna Virkkunen stated that the bloc's rulebook already covers autonomous AI threats. The regulation, which takes a risk-based approach, has been enforced in phases, with bans on certain AI practices in force since February 2025 and transparency obligations for AI providers effective as of August 2026.
Key Facts
- The EU's AI Act covers AI models across their entire life cycle, not just at launch, and includes a scientific panel of 60 experts tasked with continuously monitoring and evaluating AI models.
- General-purpose AI models trained with more than 10^25 FLOPs of compute are presumed to pose systemic risk, including loss of control, cyber offense capabilities, and manipulation at large scale.
- The European Commission's AI Office has been empowered to investigate AI models more decisively, including the authority to fine companies that fail to comply, with penalties reaching up to €35 million or 7% of global turnover.
- As of August 2026, the AI Office had sent requests to more than 30 AI providers seeking information on their safety and security practices, including transparency protocols, with high-risk systems facing deadlines of December 2027 and August 2028.