UK cybersecurity policy

The NCSC's Drop to 80 Nationally Significant Incidents Is Not a Verdict, and the Resilience Bill Should Not Treat It as One

The NCSC reports 80 nationally significant incidents, down from 204, while warning against reading a trend. Its findings point to basic hygiene, not new powers.

NCSC Annual Review 2026 at a glance People of Internet Research · UK 80 Nationally significant incidents Down from 204 in the prior year's … 1,831 Attack reports received 337 needed NCSC support after tria… 8,500 Early Warning recipients Organisations notified of suspecte… 204 Prior-year significant incidents NCSC 2025 review, up from 89. peopleofinternet.com
NCSC Annual Review 2026 at a glance People of Internet Research · UK 80 Nationally significant inci… 1,831 Attack reports received 8,500 Early Warning recipients 204 Prior-year significant inci… peopleofinternet.com

Key Takeaways

On 3 November 2026 the National Cyber Security Centre published its Annual Review 2026, The State of the Cyber Nation, covering 1 September 2025 to 31 August 2026. The headline numbers invite a victory lap: 80 nationally significant incidents, against 204 the year before. The NCSC itself declines to take one. Its message is more useful than the number, and it lands just as Parliament finishes work on the Cyber Security and Resilience Bill.

What the Review actually says

The Review records 1,831 attack reports, of which 337 needed NCSC support after triage, and 80 were classed as nationally significant. It also says 8,500 organisations received Early Warning notifications about suspected compromises (NCSC, incident management).

The prior-year comparison comes from the NCSC's 2025 review, which put nationally significant incidents at 204, up from 89 the year before, out of 429 incidents handled (NCSC, October 2025). A fall from 204 to 80 is large. But the NCSC cautions that incident numbers do not follow a simple trend, and that peaks often come from a small number of significant incidents. It also suggests Early Warning notifications likely headed off some incidents before they needed NCSC intervention.

That caveat matters for policy. One year of data cannot show that the regime is working, and it cannot show that it is failing. Anyone who cites 204 to 80 as proof of success, or the earlier rise as proof of crisis, is reading noise as signal.

The causes are familiar

The Review's more durable finding is about method. Most successful compromises still rely on familiar techniques, including stolen credentials, and many state-linked cases involved patches that existed but were not deployed in a sufficiently timely fashion. The NCSC's own summary says most successful attacks continue to exploit familiar weaknesses (NCSC, 7 things learned).

On AI, the Review is sober. Threat actors have begun using AI tools, but none of the observed uses enabled a novel class of attack. The NCSC recommends passkeys over passwords to cut credential theft, and points to the government's Cyber Resilience Pledge, which commits to rolling out Cyber Essentials across supply chains.

The AI finding should temper the political temperature. A US House report published on 20 November 2026 warns about American firms embedding Chinese open-weight models in their products (The Record). That is a supply-chain and deployment-risk argument, not evidence of a new attack class. The UK evidence so far points the same way: AI risk sits in how systems are deployed and maintained, which is where the oldest controls apply.

The Bill returns

The Cyber Security and Resilience Bill widens the organisations covered by the NIS Regulations 2018 and introduces tougher incident reporting and enforcement. The Lords completed Report Stage on 26 October 2026. They narrowed the reporting trigger to incidents that have actually affected systems, and dropped mandatory near-miss reporting for data centre operators. The Government lost two votes: Amendment 163 (202 to 155), requiring a review of the NIS regime within 12 months of Royal Assent and annually after that, and Amendment 164 (194 to 155), requiring a register of foreign powers that pose cyber threats (VinciWorks).

The case for those amendments deserves a fair hearing. State-linked activity is the NCSC's most serious concern. A public list of hostile states could give boards and procurement teams a clearer signal. Annual reviews of scope would stop the regime freezing around today's sector map while cloud and AI dependencies grow. Parliament has reason to want a statute that adapts.

But the Review supports a narrower reading. Its own evidence says the failures are mostly basic: unpatched edge devices and weak authentication. On that evidence, the Commons should prioritise three things.

Measure the right thing

The Early Warning figure is arguably the most policy-relevant number in the Review. 8,500 organisations were told of suspected compromise, and the NCSC links that service to containment before escalation. That is proportionate regulation in practice: a free, voluntary, evidence-led intervention that reduced the burden on the state and on companies. If the Bill's duties are layered on top, they should reinforce that cooperative model rather than convert it into a compliance exercise in which firms report defensively and share less.

There is also a measurement problem. The Review does not offer a definition-stable series, and it warns against trend-reading. If Parliament legislates annual reviews, it should require the Government to publish consistent methodology and denominators, so that a fall from 204 to 80 can be explained rather than merely celebrated.

The Commons will decide what to do with the Lords' changes. The sound course is to accept the reviewing mechanism, resist rigid lists, and keep the statute focused on the failures the NCSC keeps finding: patches not applied, passwords stolen, suppliers left behind. Security policy that follows the evidence will look less dramatic than a new regime, and it will probably protect more.

Sources & Citations

  1. NCSC Annual Review 2026: incident management
  2. NCSC, 2025 figures: nationally significant attacks
  3. VinciWorks: Cyber Security and Resilience Bill Lords defeats
  4. The Record: House report on Chinese open-weight AI