Skip to content
Annuaire
Sections
Cybersecurity

European Commission: what we know about the cyberattack carried out with an AI tool

L’essentiel

The European Commission confirms a cyberattack carried out with a public AI tool: HR services shut down, the ARTEX tool cited by Politico, data questions unresolved.

À retenir

The European Commission confirms a cyberattack carried out with a public AI tool: HR services shut down, the ARTEX tool cited by Politico, data questions unresolved.

The European Commission has publicly acknowledged, at a press briefing in Brussels reported on 20 November 2026 by Help Net Security, that it suffered an intrusion campaign conducted with a publicly available artificial intelligence tool, detected as early as 22 October. Several internal applications were shut down as a precaution, including the staff management tool. The institution says it has found, at this stage, no evidence of a leak affecting all of its staff, but many questions remain unanswered.

What the Commission has confirmed

It was spokesperson Thomas Regnier who confirmed the incident at the midday press briefing, a few hours after Politico’s revelations, Help Net Security reports. According to him, it was an “AI-generated” campaign based on the misuse of a publicly available tool, which he did not name. The teams spotted the suspicious activity on 22 October, then confirmed it on 26 October after an initial analysis.

The Commission says it temporarily took several applications offline as a preventive measure; most of them are working again. It is working with CERT-EU, the incident response team of the European institutions, and with the European Data Protection Supervisor. The spokesperson said he was not aware of any ransom demand and declined to comment on the attribution of the attack. Nor would he characterise the scale of the operation, which Politico described as “huge”, merely referring to a “campaign” made up of a series of incidents.

Staff services disrupted for more than ten days

Details about the affected systems come mainly from Politico, cited by The Register. Sysper, the platform where staff consult their payslips, record their working time and book their leave, was shut down. Restored a few days earlier, it went down again on Wednesday 18 November, according to The Register’s article published the following day. MyPMO, which handles reimbursements of medical expenses, was also stopped before being restored, and translation services were affected. Some employees were unable to book days off, but salaries were paid on the scheduled date.

On the question of data, accounts differ. Thomas Regnier maintains that the investigation has not uncovered any breach affecting staff as a whole. Politico, on the contrary, refers to a potentially large volume of compromised data, a contradiction that SecurityWeek notes without being able to resolve it. It is therefore still unknown whether payroll information was accessed.

ARTEX, the penetration testing tool at the centre of suspicions

The Commission has not named the tool used. According to Politico’s sources, however, its defenders spotted the use of ARTEX, an AI-driven penetration testing framework originally designed to help security teams find weaknesses in a system. On 5 November, CERT-EU had issued a warning about this tool: it said it had observed attacks targeting organisations it protects, very probably carried out with ARTEX, and considered that its free distribution would accelerate its adoption by malicious actors.

According to The Register, the developer removed the original GitHub repository in October after reports of abuse, but the code was already circulating and several derived copies remain accessible. The outlet also cites an FBI alert published the week before its 19 November article: researchers counted more than 1,600 ARTEX consoles exposed on the internet, and some attackers reportedly went from identifying a target to privileged access in under three hours. SecurityWeek mentions other victims associated with this tool, including Barracuda, Railway and ConnectWise customers.

Snehal Antani, head of the security company Horizon3.ai, sums up the stakes for The Register: attack sequences that used to take an experienced operator days or weeks now unfold in a few hours. He adds that these toolkits, shared between states and criminals, complicate attribution.

What the incident changes for organisations

The episode illustrates a shift that specialists had been predicting for months: software designed for security auditors becomes, once published, a weapon within reach of less skilled groups. The Commission’s spokesperson acknowledged this in his own way, pointing out that AI helps to spot vulnerabilities but can also be used for offensive operations, and that “no one is immune”. He referred to the action plan on cybersecurity and AI presented by the institution in July.

For a company or a public administration, the lesson is above all about pace. If the time between the discovery of a flaw and its exploitation is counted in hours, monthly patch cycles and monitoring entrusted to a single person are no longer enough. Three reflexes follow from this:

  • reduce the exposed surface (administration interfaces, HR and financial applications accessible from the internet);
  • shorten the time taken to apply patches on front-facing services;
  • plan procedures for taking systems offline quickly, as the Commission did, while measuring their cost for employees.

The communication delay is also under debate: nearly four weeks passed between the detection on 22 October and the public admission. Thomas Regnier defended this silence, explaining that measures against an ongoing campaign must remain discreet to be effective, and insisted there had been “no cover-up”.

Frequently asked questions

Was personal data of EU staff stolen?

This has not been established. The Commission says it has found no evidence of a leak affecting all of its staff, while Politico refers to potentially large quantities of compromised data. The investigation remains open.

What is ARTEX?

It is a penetration testing framework that relies on AI to automate the search for flaws. Its original repository was taken down in October, but copies are circulating. CERT-EU warned against its malicious use on 5 November; the Commission has not confirmed that it was used against it.

Who is behind the attack?

No one has been named. The Commission refuses to comment on attribution while the investigation remains open, and its spokesperson says he is not aware of any ransom demand.

Sources

Sur votre appareil

Comprendre cet article

L’analyse utilise l’intelligence locale du navigateur lorsqu’elle existe, sinon un résumé extractif. Le texte n’est envoyé à aucun service extérieur.

Facebook X LinkedIn

Ensuite A lire aussi

Free, no spam, one-click unsubscribe.