AI news
AI model reviews
AI tools
How-to guides
Our staffAboutContact

EU confirms October cyberattack used AI tools

The European Commission says an October cyberattack on the EU used AI-powered software. Some digital tools are still offline, and the investigation is ongoing.

row of european union flags
Photo: Guillaume Périgois on Unsplash

The European Commission has confirmed that a cyberattack on the EU in October used software powered by artificial intelligence, and some of the bloc's digital tools are still offline. Spokesperson Thomas Regnier said so on Thursday: "Indeed we can confirm that this was done with AI tools."

Regnier said the EU first detected the incident on Oct. 22, then opened an investigation and imposed what he called "preventative measures." "We temporarily took down several applications, and most of them have since then been restored," he said. The Commission didn't say who was behind it, and it was the first time the 27-member bloc reported a cyberattack using an AI tool.

Staff systems and the open data question

The disruption landed on internal HR systems. For several days, Commission officials couldn't access Sysper, the platform where they check salary information, request leave, submit timesheets and sign up for training. On Thursday Sysper was working again, but the health insurance scheme JSIS was still unavailable, an official told Euronews anonymously. Most civil servants barely noticed the systems had been compromised.

Data is the open question. Regnier said that so far "we have not detected any evidence of a data breach affecting commission staff as a whole." That wording is narrower than "nothing was taken", and the key question is whether the attacker, who hasn't been identified, reached the data held in Sysper, JSIS and other HR systems. I'd read the Commission's line as cautious rather than reassuring.

The tool behind it

Regnier didn't name the software. Reports built on Politico's coverage do: they say CERT-EU, the cybersecurity service for the EU institutions, found international hackers using ARTEX, a publicly available, AI-driven tool designed to identify vulnerabilities in IT systems. One summary describes it as using large language models to coordinate automated reconnaissance and security testing. Since the Commission hasn't confirmed those details on the record, I'd treat them as reported only.

The sources also disagree on timing. The Commission's account has the first detection on Oct. 22, while the Politico-based reports say CERT-EU spotted ARTEX's use earlier this month, meaning November. They may describe different stages of the same incident, but nothing I've read settles it. CERT-EU hasn't named any attackers or disclosed a specific vulnerability.

The stakes for AI companies and users

Regnier's comment points at a problem the AI industry knows well. "AI-assisted tools can be used and especially misused to detect vulnerabilities," he said. The same kind of software that helps a security team find weak spots can help an attacker do it too. CERT-EU has previously warned that AI tools can find vulnerabilities much faster and automate attacks, so attackers can move faster than organisations can patch.

As far as I can tell, none of the reporting says the EU plans new rules, or action against the makers of such tools, because of this incident. I wouldn't guess at one. If regulators do respond, publicly available security tools are probably where the argument starts, since ARTEX is described as freely accessible.

For everyday users, the reporting describes an attack on Commission staff systems, not public services, so there's nothing in it for ordinary people to act on. Honestly, the more useful signal is for organisations: if AI shortens the time between a flaw existing and someone finding it, slow patching becomes a bigger risk.

Next steps in the investigation

The investigation is ongoing, and CERT-EU is still assessing the damage. Three things are worth watching: whether the Commission finds that staff data was accessed, whether JSIS comes back, and whether anyone is formally blamed. It isn't the Commission's first recent incident either: in January, an attack on its central mobile-device infrastructure potentially exposed staff names and phone numbers.

I'd wait for the findings on data access before drawing big conclusions. The confirmation itself matters, though: an EU institution has now said publicly that AI tools were used against it.

Keep reading