ENISA's third NIS360 report, published May 28, maps cybersecurity maturity across 22 sectors under the NIS2 Directive. Electricity, telecoms and banking lead with high maturity thanks to sustained regulation and investment. However, six sectors are placed in the risk zone where maturity does not match the severity of the threat landscape: healthcare, public administration, maritime, gas, ICT service management and space.
Novudom Weekly – Week 22, 2026
This week is marked by ENISA's third NIS360 report showing six critical sectors still in the risk zone, while Verizon's DBIR 2026 confirms that supply chain involvement now accounts for nearly half of all verified data breaches. The European Commission's consultation on AI Act transparency requirements closes June 3 and demands attention from anyone developing or using AI systems.
ENISA NIS360 2026: Six critical sectors in the risk zone
Check where your sector falls in ENISA's risk matrix. Being in the risk zone means the threat level exceeds your current capability. The board needs to allocate resources to close the gap before supervision tightens further.
Verizon DBIR 2026: Vulnerabilities surpass stolen credentials as top breach entry point
This year's Data Breach Investigations Report analyses 22,000 confirmed breaches across 145 countries. For the first time in 19 years, vulnerability exploitation (31%) has surpassed stolen credentials as the most common breach entry point. Third-party involvement increased by 60% and now accounts for nearly half of all breaches, while employee use of unapproved AI tools tripled to 45%.
Three priorities for the board agenda: strengthen vulnerability management, tighten third-party oversight, and establish a clear AI tools policy before shadow usage creates regulatory problems.
GitHub breach: 3,800 internal repos stolen via poisoned VS Code extension
Threat group TeamPCP compromised the VS Code extension Nx Console for 18 minutes on May 18, distributing a credential stealer that harvested keys from 1Password, GitHub, AWS and Anthropic configurations. An employee installed the trojanised version, giving the attackers access to 3,800 internal repositories. GitHub states that customer data was not affected.
18 minutes of exposure was enough for a full-scale breach. Developer tools are part of your attack surface. Require approval processes for extensions in the development environment, and inventory which tools are actually in use.
Carnival breach: 6 million affected after social engineering attack
Cruise giant Carnival confirmed on May 27 that a breach affected 5.9 million individuals. Attackers used social engineering against an employee to gain initial access on April 14. Exposed data includes names, addresses, dates of birth, email addresses and in some cases passport and ID numbers. Formal notification letters were sent only six weeks after discovery.
Social engineering bypasses technical controls. A six-week delay between discovery and notification raises questions about incident response processes. Review your own readiness: how quickly can you identify, contain and communicate a breach?
AI Act: Transparency guidelines open for consultation, deadline June 3
The European Commission opened a consultation on May 8 on draft guidelines for the transparency requirements under AI Act Article 50. The guidelines specify how providers must machine-readably mark synthetic content, inform users about AI interaction, and handle deepfakes. The transparency rules take effect on August 2, 2026.
The consultation closes June 3. If your organisation develops, provides or uses AI systems that generate text, images or audio: map which transparency requirements apply to you now.
Italy's Garante: Email tracking pixels require explicit consent
Italy's data protection authority ruled that tracking pixels in emails are only permitted if the recipient has been informed in advance and, in most cases, has given explicit consent. At the same time, the Garante classified all communications in individually assigned corporate email accounts as personal data, significantly expanding employee privacy protection.
Two rulings to watch: if you use email tracking in marketing or internal communication, ensure you have a legal basis. The classification of corporate email as personal data affects data handling, archiving and e-discovery procedures.
Iranian state attack: MuddyWater hits nine organisations in nine countries
Security researchers link a new campaign by the Iranian group MuddyWater to attacks against at least nine organisations in nine countries during Q1 2026. Targets include manufacturing, education, public sector, financial services and professional services. The campaign shows that state-backed actors are broadening their targets beyond traditional geopolitics.
State-sponsored threats are not reserved for defence and government. If your organisation operates in manufacturing, education, finance or professional services, your threat assessment should include state actors.
Weekly reflection
This week's reports point in the same direction: the attack surface is growing faster than defensive capability. ENISA's NIS360 shows six sectors lagging behind. Verizon's DBIR confirms that the supply chain is now involved in nearly half of all breaches. And the GitHub incident proves that 18 minutes is enough. The question for every board this week is not whether the risk is high, but whether the organisation can actually demonstrate the resilience that NIS2 and the AI Act require before the next supervisory round.