← All issues
Week 28 · 2026

Novudom Weekly – Week 28, 2026

EU launches AI cybersecurity action plan, first AI-driven ransomware confirmed, and Accenture suffers major breach

Week 28 shows the line between AI regulation and cybersecurity is blurring fast. The European Commission presented its first action plan for AI and cybersecurity the same week the first fully AI-driven ransomware attack was documented. Meanwhile, the Accenture breach reveals that not even the largest consultancies are immune, and the CRA reporting deadline approaches with a market largely unprepared.

AI Act

EU launches action plan: AI to be tested before market entry

On July 7, the European Commission presented its action plan for cybersecurity and artificial intelligence. The plan addresses how advanced AI models can both strengthen and undermine cybersecurity. Key elements include requirements for independent evaluation of AI models before EU market entry, a European blueprint for structured access to advanced AI through ENISA, and a joint testing platform for AI in cybersecurity environments with JRC.

What it means for you

Organizations using or delivering AI-based services should expect third-party assessment requirements. AI governance and cybersecurity are now treated as one and the same issue by the EU.

Source: EU-kommissionen →
Cyber

Accenture confirms breach: 35 GB of source code and cloud keys stolen

A threat actor using the alias 888 has listed what is claimed to be 35 GB of stolen data from Accenture for sale on PwnForums. The material allegedly includes source code, RSA keys, SSH keys, Azure access tokens, and configuration files. Accenture confirms an isolated security incident and states operations are unaffected, but has not commented on the scope or whether customer data was exposed.

What it means for you

Stolen cloud keys and access tokens can give attackers access to customer infrastructure across the supply chain. Ask your consulting providers about key rotation and scope limitation.

Source: Help Net Security →
Cyber

CRA: 63 days to reporting deadline, two thirds of the market unprepared

On July 2, ENISA published an FAQ on its Single Reporting Platform (SRP), which becomes the mandatory reporting channel for actively exploited vulnerabilities and severe incidents from September 11. The 2026 CRA Awareness and Readiness Report shows that 66 percent of surveyed organizations still do not know about the regulation, up from 62 percent in 2025. Manufacturers of products with digital elements must report within 24 hours of discovery.

What it means for you

If you manufacture or sell products with software in the EU, you have 63 days left to establish vulnerability reporting processes via ENISA's platform. The readiness gap is alarmingly wide.

Source: Crowell & Moring →
NIS2

NIS2: 24 countries live, Netherlands activates, no fines yet

With the Netherlands' Cyberbeveiligingswet entering into force around July 1, 24 of 27 EU member states have now transposed NIS2 into national law. France and Spain remain in active legislative processes. Despite supervisory regimes being operational in most countries, no national authority has yet issued a finalized administrative fine under NIS2. Legal experts describe 2026 as the year when the first enforcement actions begin.

What it means for you

The absence of fines should not be interpreted as breathing room. Supervision is live and already reviewing. Organizations in essential sectors should view this as a grace period, not a sign of passivity.

Source: Skadden →
Cyber

JadePuffer: First fully AI-driven ransomware attack documented

Security researchers have documented what is described as the first complete ransomware attack driven by large language models (LLMs). The attack chain, dubbed JadePuffer, uses AI throughout the entire attack flow: from initial reconnaissance and social engineering through lateral movement to encryption and extortion. The incident was confirmed on July 6 and marks a shift from AI as an aid to AI as an autonomous attacker.

What it means for you

AI-driven ransomware dramatically lowers the barrier for attacks. Boards should address whether their organization's detection capabilities are sufficient for attacks that adapt in real time.

Source: Dark Reading →
Styrning

D&O insurance: Cyber incidents reshape board members' risk exposure

The Directors & Officers insurance market is undergoing rapid transformation in 2026 as cyber incidents drive more lawsuits and regulatory actions. The Caremark standard is now being actively applied to board cybersecurity oversight: directors risk personal liability if they lack reporting systems for cyber risk or consciously fail to monitor existing systems. Several insurers have already raised premiums and introduced requirements for documented cyber governance.

What it means for you

Board members who cannot demonstrate documented oversight of cyber risk expose themselves personally, both legally and in their D&O insurance. Documented board treatment of cyber issues is not just good practice but an insurance requirement.

Source: Governance Intelligence →
AI Act

ENISA: New report on AI and cybersecurity with recommendations for authorities

On July 7, ENISA published its report Artificial Intelligence and Next Gen Technologies with recommendations aimed at national supervisory authorities, policymakers, and service providers. The report analyzes how AI and emerging technologies are changing the threat landscape and provides concrete guidelines for how authorities can build capacity to handle AI-related cyber threats. ENISA also launched a survey for NIS2 entities as input for the next edition of the NIS360 report.

What it means for you

ENISA's report provides a picture of what supervisory authorities themselves are preparing for. Organizations in NIS2 sectors would do well to read it as an indicator of upcoming supervisory focus.

Source: ENISA →

Weekly reflection

This week the question was put to the test: is AI a tool for defence or for attack? The answer is both, and that requires an entirely new governance model. The EU's action plan and the JadePuffer attack are two sides of the same coin. Organizations that treat AI governance and cybersecurity as separate issues will find themselves on the wrong side of the curve. This is no longer a technology question. It is a leadership question that determines which organizations will manage the regulatory and operational transition, and which will not.

Want to know where your organisation stands? A NovuLens analysis gives you the answer in under a week.
Contact us