ENISA expects AI to support more and more cyberattacks in the EU

On 22 September, ENISA, the European Union's cybersecurity agency, published its annual threat report. It analyses 8,257 incidents that hit the EU in 2025.

Denial-of-service attacks (51.3%) and unauthorised access (39.5%) made up most of them. Public administration was the most affected sector (31.8%).

On AI, the report says it kept being integrated into malicious activity in 2025, by criminal groups, state-linked actors and information-manipulation operations. ENISA assesses it is highly likely that AI will increasingly support these operations, and likely that more phases of an attack will be AI-enabled in 2026, with possible human-out-of-the-loop proofs of concept.

These are the agency's assessments, not measurements.