On May 7, the European Parliament and the Council reached a provisional agreement on amendments to the AI Act. High-risk AI systems get an extra year until December 2027. An explicit ban on AI apps generating sexual content without consent is also introduced. The agreement must be formally adopted before August 2, 2026.
Novudom Weekly – Week 19, 2026
This week is dominated by the EU's political agreement on a simplified AI Act, new reports from ENISA, WEF and Europol showing how the threat landscape is shifting towards AI-driven fraud, and the increasingly clear personal liability for boards and management under NIS2.
AI Act Omnibus: EU agrees on simplification and ban on nudification apps
Management gets breathing room for high-risk AI compliance but should not delay. The nudification ban signals that the EU takes AI misuse seriously, affecting all organisations developing or using generative AI.
Europol IOCTA 2026: Cybercrime industrialises with AI and ransomware
Europol's Internet Organised Crime Threat Assessment 2026 shows that cybercrime in the EU has industrialised. Over 120 active ransomware brands were observed in 2025. AI is increasingly used to automate phishing, generate deepfakes and lower the barrier for new actors. Extortion methods are diversifying with DDoS attacks and cold-calling of victims.
The threat landscape requires the board to understand that ransomware is no longer an isolated IT risk but an industrialised business model. Incident preparedness and supplier monitoring should be standing agenda items.
NIS2 Article 20: Personal liability for boards and CEOs becomes reality
As 21 of 27 EU countries have now transposed NIS2 into national law, focus sharpens on Article 20, which requires management bodies to personally approve and oversee cybersecurity measures. Sanctions can include temporary suspension from management duties, civil damages and fines up to EUR 10 million or 2 percent of global turnover.
Board members and CEOs can no longer delegate cybersecurity to the IT department. Documented board involvement in risk assessments and action plans is now a legal necessity.
ENISA 2026 Risk Landscape: DDoS tops and attackers hide in legitimate services
ENISA's latest risk report shows that availability attacks (DDoS) and ransomware continue to dominate the EU threat landscape. A new trend is Living Off Trusted Sites (LOTS), where attackers hide command and control in ordinary web traffic via legitimate platforms. Public sector is increasingly targeted by hacktivist groups.
Traditional perimeter defences are insufficient when attackers exploit trusted platforms. Organisations should invest in behaviour-based detection and ensure the incident reporting chain works according to NIS2's 24-hour requirement.
WEF: AI fraud tops CEO threat list, geopolitics reshapes risk landscape
The World Economic Forum's Global Cybersecurity Outlook 2026, based on 804 respondents in 92 countries, shows that phishing and AI-driven fraud have now surpassed ransomware as the top cyber threat. 94 percent of respondents identify AI as the most significant change driver. The share of organisations actively evaluating the security of their AI tools has nearly doubled to 64 percent.
Fraud-related cyber attacks require different countermeasures than traditional technical defences. Management should prioritise training, process validation and AI governance alongside technical security.
DORA Register of Information: First submission reveals compliance gaps
The first annual submission of DORA's Register of Information was completed in March 2026. Results show that nearly half of all financial institutions struggle to fully map their ICT subcontractors. Common gaps include incomplete subcontractor mapping and inconsistent provider identification. Supervisory authorities now cross-reference data automatically.
The financial sector should treat the Register of Information as a living document, not an annual exercise. Boards need to ensure third-party risk management has adequate resources and mandate.
European Commission proposes targeted NIS2 reform for increased clarity
On January 20, 2026, the European Commission proposed targeted amendments to the NIS2 directive. The goal is to increase legal certainty, simplify compliance and reduce the administrative burden for organisations subject to multiple overlapping EU regulations. The proposals are particularly relevant for organisations simultaneously affected by DORA, the CER directive and the AI Act.
Organisations navigating multiple EU regulations should follow the reform work closely. A consolidated compliance strategy covering NIS2, DORA and the AI Act in parallel saves both time and resources.
Weekly reflection
As the EU's AI Act is simplified and the NIS2 reform takes shape, a pattern emerges: regulatory frameworks are converging. Organisations building separate compliance silos for each directive risk duplicating work while still missing the bigger picture. The most effective strategy going forward is an integrated governance model that addresses NIS2, DORA, the AI Act and GDPR through shared processes for risk assessment, incident management and supplier oversight.