Europe's AI disinfo rules aren't built for agentic attacks

In 2025, one in four foreign disinformation operations detected by the European External Action Service involved AI tools, and that figure is the frame for a new argument in TechPolicy.Press from Cambridge researcher Gerald Mako. His claim is not that Europe has ignored AI-enabled interference. It is that the instruments Europe has built for it, the Digital Services Act, the AI Act, and the European Democracy Shield, are aimed at the wrong layer of the problem.

The layer they miss is agentic. Mako's description of the new systems is worth reading closely: they are goal-oriented, and once given an objective they will figure out how to achieve it by navigating platforms, using external tools, checking their own progress, and pivoting when necessary, all with barely any human involvement. A content moderator hunting a synthetic image or a suspicious post is looking at a single artifact. What Mako is describing is a network of artifacts produced under a shared objective, coordinated in a way current compliance-driven rules were not written to see.

Russia, in Mako's telling, has already demonstrated the earlier version of this playbook through campaigns such as Doppelgänger and Operation Overload, and Russian networks have reportedly flooded the open web with millions of low-quality articles specifically designed to be scraped into training data. That second move is the slower one, seeding the raw material future models learn from rather than any single vote.

The honest caveat is that this is an op-ed from one researcher affiliated with the Cambridge Central Asia Forum, not a fresh intelligence disclosure. It leans on the EEAS's one-in-four figure and on named campaigns already in the public record. What the piece does not give you is a technical breakdown of how coordinated-agent detection would actually work at platform scale, or how much of Russia's agentic capability is already fielded versus projected from adjacent tradecraft.

Where it does push the conversation forward is on tooling. Mako argues for permanent red-teaming operations for Foreign Information Manipulation and Interference, and for a dedicated funding line for European public-interest AI tooling, focused on detection of coordinated agent behavior and rapid counter-narrative generation. The bet is that Europe cannot regulate its way out of an operations problem, and that the sovereign-tooling question is now the one worth budgeting for.