European Commission Vice-President and Digital Technology Commissioner Henna Virkkunen has issued a stern warning to AI industry leaders, urging them to maintain absolute control over their systems. Speaking at the AI Summit in Dublin, the Finnish official argued that the relentless race between tech giants is leading to the premature release of models without adequate safety testing.
Losing Sight of AI Agents
A primary concern for Brussels is the evolution of autonomous AI agents. Virkkunen noted that global competition is now so intense that companies appear to be neglecting the necessary time and resources for thorough evaluation. "We have seen incidents revealing that companies do not always have knowledge of the actions their agents take," she stated, emphasizing that human oversight must remain a priority.
As AI models grow more powerful, their evaluation becomes increasingly complex. Virkkunen stressed that independent experts must be granted sufficient time to conduct rigorous safety checks, a process currently threatened by market pressures.
Enforcement via the AI Act
The European Union has begun exercising its powers under the pioneering AI Act. Commission services have already dispatched formal Requests for Information (RFIs) to key players in the sector, seeking detailed explanations on several fronts.
The inquiries focus specifically on:
- Internal safety measures currently in place.
- External evaluation procedures, including red-teaming.
- The protection of intellectual property rights.
According to the Commissioner, transparency and substantive control of these systems are fundamental requirements of the new European regulatory framework, not mere suggestions.
