Skip to content

Response, New York. Noah Park: Check Point Patches Actively Exploited Management Server Path Traversal. Detection, Dallas. Ava Okello: Report Finds Nearly Half of Deployed Detections Need Attention Before SOCs Can Trust Them.

SOCtember

Always First. Fast SOC News.

Screenshot of the European Commission Audiovisual Service page for the EC Midday press briefing of 19/11/2026, showing spokesman Thomas Regnier at the podium in the Berlaymont press room.
Image: European Commission Audiovisual Service (fair use). Thursday's midday briefing, where the Commission answered questions on the cyber campaign.

Response

European Commission Confirms an AI-Generated Cyber Campaign It Detected Oct. 22, After Weeks of Staff Systems Going Dark With Little Explanation

The Commission's spokesman said a publicly available AI tool was misused, several applications were taken offline as a precaution and no evidence has been found of a breach affecting staff as a whole. POLITICO reported that CERT-EU identified the tool as ARTEX, the open-source attack framework CERT-EU had warned its constituency about on Nov. 5.

Noah Park, ResponseNew York6 min read

NEW YORK - The European Commission confirmed on Thursday that it had been the target of what it called an "AI generated cyber campaign," first detected nearly a month ago, after a news report described weeks of disruption to the personnel systems its staff use for payslips, leave and medical claims.

Speaking at the Commission's midday press briefing in Brussels, the spokesman Thomas Regnier said the incident "was detected on the 22nd of October and it was confirmed following the initial investigation on the 26th of October." As "a protective and preventive measure," he said, "we temporarily took down several applications, and most of them have since then been restored."

Mr. Regnier said the Commission was in close contact with CERT-EU, the cybersecurity service for the European Union's institutions, and with the European Data Protection Supervisor, and that the investigation was continuing. "We have not detected any evidence of a data breach affecting Commission staff as a whole," he said.

A publicly available tool

Asked whether AI tools were part of the intrusion, Mr. Regnier said the Commission could "confirm that this was an AI generated cyber campaign," and described it as "a misuse of an AI tool." Pressed on how the attackers got in, he said "it was an AI tool that is publicly available," without naming it. He declined to discuss attribution, the attackers' goals or the tools the Commission used in its defense, said he was "not aware of any ransomware request made to the Commission," and would not characterize the scale of the incident. "It is a campaign, so there was a series of incidents indeed," he said.

The confirmation followed an exclusive report early Thursday in POLITICO's Brussels Playbook newsletter, which said CERT-EU had detected this month that the attackers used ARTEX, which the newsletter described as a publicly available AI-powered tool designed to identify vulnerabilities in IT systems.

Screenshot of the POLITICO Brussels Playbook newsletter titled Cyberattack hits the Commission, by Gerardo Fortuna, dated November 19, 2026, 6:26 a.m. CET.
Image: POLITICO (fair use).The Brussels Playbook report that preceded Thursday's confirmation.

Ten days of disruption

According to POLITICO, the disruption lasted more than 10 days. Sysper, the Commission's human resources platform for payslips, leave requests and timesheets, and MyPMO, its medical reimbursement system, were restored earlier this week, the newsletter reported, though Sysper briefly crashed again on Wednesday. Translation services were hit on Tuesday.

"The Commission believes it has contained most of the attack, but the full extent of any damage remains unclear," an EU official told POLITICO on condition of anonymity. The newsletter said the Commission's top internal crisis body, the Corporate Management Board, held an urgent meeting this month over "ongoing cyber incidents," citing a leaked internal email, and that Henna Virkkunen, the commissioner for technology, briefed fellow commissioners on Wednesday.

Staff were told little. POLITICO said salaries were paid as usual, but some officials could not file timesheets or request leave, and an IT reply to one Sysper complaint cited a general problem with the system without mentioning a cyberattack.

Asked by a German broadcaster why the Commission had not disclosed the attack itself, Mr. Regnier said measures taken during a cyber campaign are kept private so they remain effective, that affected services and the data protection supervisor had been informed, and that "there was no hiding whatsoever."

CERT-EU's warning

Two weeks before the report, on Nov. 5, CERT-EU published a warning that several threat actors around the world were using ARTEX to automate their campaigns and that it had "observed similar instances of AI-driven activities against its constituency," which it said "highly likely involved the use of ARTEX." It did not name a victim.

Screenshot of the CERT-EU blog post From Penetration Testing to Cyber Threat: Beware of ARTEX, published November 5, 2026, describing ARTEX as an AI-powered penetration-testing framework and noting AI-driven activity against CERT-EU's constituency.
Image: CERT-EU (fair use).CERT-EU's Nov. 5 warning about ARTEX.

CERT-EU described ARTEX as an LLM-driven autonomous penetration testing platform that takes an objective in natural language and has agent workers carry out reconnaissance and testing with real tools, steered from a dashboard by a human operator. It relies on the model and on configurable rules, not hard technical controls, to stay in scope. Its original author made the repository unavailable in October after reports of misuse, CERT-EU said, so the versions in circulation are independent forks that plug into whatever external AI model an operator supplies.

What makes the tool dangerous, CERT-EU wrote, is scale, speed and persistence: it can test large numbers of low-complexity misconfigurations that a human tester would notice but not have time to check, and work through long attack chains without fatigue.

What response teams can take from it

CERT-EU's guidance, published before the Commission incident became public, is aimed at exactly this pattern. Its immediate steps are to identify internet-exposed assets and cut exposure to those that need it, isolate management and control panels from the internet, prioritize emergency patching of anything exposed, and closely monitor privileged and administrator accounts for anomalous activity.

Screenshot of the What you can do section of the CERT-EU ARTEX post, listing immediate, short-term and continuous measures including limiting internet exposure, isolating management panels, monitoring privileged accounts, MFA, hunting for ARTEX artefacts and sanitizing error responses.
Image: CERT-EU (fair use).CERT-EU's recommended measures.

In the short term, it recommends multifactor authentication on all critical systems, hunting for connections to known actor infrastructure and for artifacts consistent with ARTEX, such as automated scanning patterns and tool signatures, in an organization's logs, and using endpoint detection and response tools to block processes that attempt privilege escalation. Its continuing measures include sanitizing error responses from internet-facing devices to make fingerprinting harder and keeping even low-severity findings in the remediation cycle, so an AI tool cannot chain minor misconfigurations into a serious compromise.

The Commission's incident is its third disclosed this year, after breaches involving Ivanti software in January and its cloud infrastructure in March, The Record from Recorded Future News reported. Officials have not linked the three.

About the Author

Noah Park

Noah Park covers incident response, containment, and blue-team operations for SOCtember from New York.

Related stories

Response desk