Picture a large European energy company that finds intruders in its network and has to pull its payroll system offline. Under the EU's NIS2 directive it has 24 hours to send an early warning to the national cyber authority, 72 hours to file a proper notification and a month to deliver a final report. If personal data is at high risk, the GDPR says the people affected must be told without undue delay. Get it wrong and NIS2 allows fines of up to €10 million or 2% of global turnover. Since September, under the Cyber Resilience Act, manufacturers selling software or connected products in Europe have had 24 hours to flag an actively exploited flaw, with fines of up to €15 million.
Now picture the institution that wrote those rules.
On Thursday morning Politico's Brussels Playbook reported that a "huge" cyberattack had hit the European Commission. By lunchtime, the Commission's digital spokesperson, Thomas Regnier, was confirming it at the daily press briefing. The attack, he said, was detected on 22 October and confirmed on 26 October. That is four weeks between detection and the first public word, and it came only after Politico had the story.
What we actually know
Not much, which is rather the point. Regnier described a "cyber campaign", which he explained meant a "series of incidents", Rapporteur reports. He confirmed it was carried out with a "publicly available" AI tool. According to Politico, the EU institutions' own cyber team, CERT-EU, found the attackers using ARTEX, an AI-powered tool built to hunt for weaknesses in IT systems. Regnier said he was not aware of any ransom demand. On who did it, he would "not go there". Rapporteur adds that he could not say whether the attacks were over.
What we know of the damage comes mostly from leaks. The Commission took several internal applications offline as a precaution. They included Sysper, the HR platform staff use for payslips, leave requests and timesheets, and MyPMO, the medical reimbursement system, Politico reports. Disruption lasted more than ten days. Translation services were hit too. On Thursday the staff health insurance scheme was still unavailable, two officials told Euronews. The Commission's top internal crisis body, the Corporate Management Board, held an urgent meeting, and tech commissioner Henna Virkkunen briefed her colleagues on Wednesday, according to Politico.
And the data? "We have not detected any evidence of a data breach affecting Commission staff as a whole," Regnier said.
Two words doing heavy lifting
Read that sentence twice. "As a whole" is the sort of phrase that survives a lawyer's red pen precisely because it rules out so little. It does not say no staff data was taken. It says the Commission has not found evidence that everybody's was.
That matters, because of what sits in these systems. The HR platforms hold names, pay grades, home addresses and contact details, Euronews reports. The Commission employs more than 33,000 people, according to Tagesschau, citing the Commission's own website. A list of who works where in the EU's executive, what grade they hold and where they sleep at night is not a spreadsheet. It is a targeting pack for phishing, pressure and worse, and Euronews notes it could fuel far more sophisticated campaigns against individual officials.
So how were those officials told? Regnier told Politico that "in line with our standard practice, we have informed those concerned". But an official told Politico that communication to staff had been limited because nobody yet knows what was taken, and Euronews reports that officials have yet to receive any official internal communication about the incident. Brussels' own cyber rulebook for EU institutions, Regulation 2023/2841, tells them to tell the users of affected systems, "without undue delay", what they can do to protect themselves. It is fair to ask how much delay a month has to be before it becomes undue.
The fair objection
There is a defence, and it deserves a hearing. Investigators often keep quiet to avoid tipping off attackers who may still be inside. No organisation is immune: on the same day, Germany's BSI raised its cyber threat level to "tense to critical", warning that AI now lets attackers exploit weaknesses at "previously unknown volumes and speeds", Tagesschau reports. The Commission says it has informed CERT-EU and the European Data Protection Supervisor, which is what the rules require.
All true. It is also exactly what every hacked company says. And Brussels' laws do not let companies hide behind it. The reporting clocks in NIS2 and the Cyber Resilience Act start ticking on awareness, not on certainty.
The Commission also knows how to do better, because it has done it. When its europa.eu cloud hosting was breached in March, it went public within days, and CERT-EU later published a detailed technical account: roughly 91.7 GB of data taken, including names, email addresses and email content, with at least 29 other EU bodies potentially affected. In January, Euronews notes, another attack hit the Commission's mobile-device infrastructure. That makes this at least the third publicly known attack on the Commission this year. Practice should be making the Commission better at talking, not quieter.
Regnier's own reassurance on Thursday was that the Commission presented an AI and cybersecurity action plan in July, which "shows that we are very well aware of what's going on around us". Its staff might settle for being aware of what is going on in their payroll system.
One rule for them
Here is the part that should annoy every European business. Under the GDPR, a company that fails to notify a data breach properly faces fines of up to €10 million or 2% of turnover. EU institutions are covered by their own data-protection regulation. Under its Article 66, the same failure can cost them at most €25,000 per infringement and €250,000 a year, and only after they have ignored an order from the data-protection supervisor. The rule-maker wrote itself a discount of roughly 99.75% on the headline penalty.
To be fair again, the Commission has noticed that its rules are heavy. Exactly one year ago today, in its Digital Omnibus, it proposed letting companies report a data breach to regulators only when it poses a high risk, and giving them 96 hours instead of 72. That was the right instinct. Perhaps the past four weeks will give Brussels some sympathy for the firms that live under its clocks.
The Dutch liberal MEP Bart Groothuis put the political cost plainly to Euronews: each time the Commission keeps intrusions quiet, he said, it misses a chance to "explain and sell" Europe's economic security policy.
What should happen now
None of this requires a new directive. It requires the Commission to do three things it demands of everyone else. Publish a timeline: when the attackers got in, when they were found, which systems they touched. Tell every member of staff, in writing, what Sysper and the health scheme hold about them and what to watch for. And publish a summary of CERT-EU's findings, as it did in March, when the investigation allows.
Then, while it is at it, fix the double standard. Either EU institutions face the same consequences as the companies they regulate, or companies get the same leniency as the institutions. Brussels can choose. It cannot keep both.
We noted this month that Brussels' flagship cyber law does not even cover the supermarket whose freezers were hacked. Now we learn that the rule-maker's own penalties are far gentler than the ones it sets for the supermarket's owner. The Commission wants to be the world's digital rule-maker. Rule-makers lead by example, and the first rule of a breach is the one it wrote for everyone else: say what happened, and say it fast.