The European Commission, the executive branch of the European Union, was targeted by a cyberattack in October that was carried out using artificial intelligence and disrupted internal services for more than ten days, an official confirmed this week. The Associated Press reported that the attack against the EU executive was executed with the help of an AI tool, in one of the most consequential publicly disclosed uses of AI in an attack against a government institution to date.

What happened For more context on this story, see our ongoing AI industry coverage.

According to the Associated Press, the attack took place in October and used a publicly available AI tool known as ARTEX. The intrusion forced the Commission to take internal applications offline, including the human resources and payroll systems that serve its tens of thousands of staff. Techzine reported that the cyberattack shut down the Commission's HR systems, while the outlet Hromadske and Spain's Escudo Digital both reported that internal services remained affected for more than ten days.

The Commission says it has contained most of the intrusion, Escudo Digital reported, but investigators are still working to determine whether personal data belonging to employees was compromised. Euronews described the incident as a large-scale cyberattack with the extent of any data leakage still unknown, and Digital Watch Observatory reported that the attack may have exposed substantial amounts of personal data. No attacker has been publicly named, and the Commission has not attributed the operation to any state or group.

Why the ARTEX detail matters

What sets the incident apart is not the target — EU institutions are routinely probed — but the tooling. ARTEX is a publicly available AI tool, and its use shows how quickly offensive capabilities that once required elite teams are diffusing into broadly accessible software. CyberSecurityNews reported that hackers targeted the European Commission with the ARTEX AI tool in what it characterized as a major cyberattack, and the Times of India noted that the disclosure lands just as the FBI has been warning about AI-assisted cyber operations, with the Bureau's warnings about AI tools newly in focus after the breach.

For the EU, the symbolism is uncomfortable. The bloc has spent years building one of the world's most extensive regulatory frameworks for digital technology, from the GDPR to the AI Act, and has positioned itself as the world's rule-setter for artificial intelligence. That one of its own core institutions was disrupted for ten days by an attack leveraging a public AI tool underscores the gap between regulating AI and defending against its misuse.

The investigation and open questions

Several critical questions remain unanswered. The Commission has not said how the attackers gained initial access, whether the ARTEX tool was used to craft the intrusion itself, to automate reconnaissance, or to generate persuasive lures — only that the attack was executed with its help. It has also not confirmed whether employee personal data actually left its networks, an assessment Euronews reported is still underway.

The duration of the disruption is notable in itself. Payroll and HR platforms are operationally critical for an institution that employs tens of thousands of people across Brussels and dozens of national offices, and keeping them offline for more than ten days suggests either a sustained effort to evict intruders or a deliberate quarantine to prevent lateral movement — a decision that prioritizes containment over continuity. Romania-based outlet Informat.ro reported that internal applications were only temporarily disabled, which is consistent with a staged restoration in which systems are brought back gradually as they are verified clean.

The disclosure also arrives amid a broader stretch of AI-related instability in critical systems. Ukrainian media reported this week that artificial intelligence may have been indirectly involved in a series of large-scale IT outages in the United States, and separate reporting has linked AI-discovered vulnerabilities to recent patching disruptions — a pattern that has made AI-driven operations a recurring theme in security briefings on both sides of the Atlantic.

A warning shot for institutions everywhere

Security analysts have warned for the past two years that generative AI would lower the barrier to sophisticated operations — automating vulnerability research, drafting convincing phishing at scale, and accelerating every phase of an intrusion. The Commission incident is the clearest public confirmation yet that this is happening against high-value government targets, using tools anyone can download. The EU's own cybersecurity agency framework, built around ENISA and the NIS2 directive on critical infrastructure security, now faces the kind of test regulators had mostly written about rather than experienced firsthand.

The response so far has been containment, not retaliation or naming. As the investigation continues, the case is likely to feature in the escalating policy debate on both sides of the Atlantic about AI-enabled attacks on government systems, and about whether public availability of powerful AI tooling needs the kind of guardrails regulators have so far focused on applying to commercial models rather than to attackers.

---

Stay Ahead of AI

Get the latest AI news, analysis, and breakthroughs — all in one place.

Read more AI news →