What counts as a significant incident under the Cyber Security and Resilience Bill?
Under the Cyber Security and Resilience Bill (CSRB), you would have to report an incident that has actually affected the operation or security of the systems behind your service, where the UK impact is or is likely to be significant. The Lords narrowed the definition at Report Stage, so blocked attempts and near misses are not reportable.
Last updated · Precursor SecurityStatus checked : the Bill passed the House of Lords on 2 November 2026 and returns to the Commons to consider Lords amendments (no date set). On 4 November 2026 the Commons extended its carry-over, so it lapses if not passed by the end of 29 April 2027. It is not yet law. References are to HL Bill 62 (as amended on Report), so wording and numbering may still change in the final Act. This guide is general information, not legal advice.
Key points
- HL Bill 62 defines an incident as an event with an actual adverse effect on the operation or security of systems.
- Under HL Bill 62, OES, RDSPs and RMSPs would judge significance against the same seven factors; data centres have their own test.
- Mandatory near-miss reporting was removed at Report Stage; the Government has promised a voluntary route instead.
- There are no numeric thresholds in the Bill, so set and document your own internal triggers.
What is a significant incident?
A significant incident is one you would have to report to your regulator. Under the Cyber Security and Resilience Bill (CSRB), as amended on Report (HL Bill 62), most regulated organisations would have to report an incident that has actually affected the operation or security of the systems behind their service, where the UK impact is, or is likely to be, significant. The Bill sets factors rather than numbers.
For OES, RDSPs and RMSPs, both of these must be true:
- Effect: the incident has affected, or is affecting, the operation or security of the network and information systems relied on to provide the service.
- Impact: the impact in the UK, or any part of it, has been, is or is likely to be significant, having regard to the listed factors.
What counts as an "incident" now?
Only an event with an actual adverse effect. The definition of an incident changed twice during the Bill's passage, and the final Lords text went back to the narrower version. Our guide to what Parliament changed covers the other Report Stage amendments.
| Version | Definition of "incident" in regulation 1(2) |
|---|---|
| NIS Regulations 2018 (current law) | Any event having an actual adverse effect on the security of network and information systems |
| HL Bill 32 (June 2026) | Any event having, or capable of having, an adverse effect on the operation or security of those systems |
| HL Bill 62 (as amended on Report) | Any event having an actual adverse effect on the operation or security of those systems |
At Report Stage, the minister, Baroness Lloyd of Effra, said: "The Bill now makes it clear that an incident is an event which has had an actual adverse impact on the security and operations of an entity's systems." She added that the revised definition still covers ransomware and "pre-positioning attacks that have successfully compromised a security system".
So "operation" is new, but "capable of having" has gone. A blocked attack with no effect on your systems isn't an incident under the Bill.
How does the test differ by category?
Not much now, except for data centres. A Government amendment at Report gave operators of essential services the same seven factors as digital and managed service providers.
| Category | Regulation | Test |
|---|---|---|
| Operator of essential services | 11 | Has affected or is affecting systems, and impact is or is likely to be significant |
| Relevant digital service provider | 12A | Same test and factors |
| Relevant managed service provider | 14E | Same test and factors |
| Data centre service | 11A | An incident which has had, is having or is likely to have a significant impact on the operation or security of the systems behind the service, including through its effect on customers' systems |
HL Bill 32 also covered data centre incidents that "could have had" a significant impact. The Government removed that wording at Report, along with mandatory near-miss reporting for data centres.
Which factors decide whether impact is significant?
For OES, RDSPs and RMSPs, the factors are:
- The extent of disruption to the service, actual or likely.
- The number of users affected, actual or likely.
- How long the incident lasts.
- The geographical area affected, actual or likely.
- Whether the confidentiality, authenticity, integrity or availability of data relating to the service (the essential service, relevant digital service or managed service) has been, is being or is likely to be compromised. HL Bill 32 said data relating to users of the service; Grand Committee widened it.
- Any impact on the network and information systems of users of the service.
- Any impact on the economy or the day-to-day functioning of society.
Each factor includes likely effects. If a compromise has already happened and significant impact is likely, you shouldn't wait for the harm to appear.
Are there numeric thresholds?
Not in the Bill. Today, NIS guidance for digital service providers, first published by the ICO and now the Information Commission's, uses numeric thresholds, such as service unavailability over 750,000 user-hours or a loss of integrity or confidentiality affecting more than 15,000 UK users. That regime rests on retained EU Implementing Regulation 2018/151.
The Bill revokes that Implementing Regulation in Schedule 2, paragraph 17 of HL Bill 62, and drops its definition from the NIS Regulations. Unless regulators publish new figures in guidance, the test becomes a judgement against the factors. In our view, that makes your own documented thresholds more important.
Why did the Government change the test?
Because almost nothing was being reported. DSIT's impact assessment, citing the 2022 post-implementation review, said only 13, 12 and 22 NIS incidents were reported in 2019, 2020 and 2021. It called the old definition "too narrow".
DSIT modelled 2 to 16 reports a year for each newly regulated organisation, with 7 as the central case. Those figures were produced before the Report Stage narrowing, so treat them as a rough guide only.
Worked examples
These are our reading of the text, not regulator decisions. Real cases depend on the facts.
| Scenario | Category | Likely position | Why |
|---|---|---|---|
| Ransomware stops a managed service for many customers for hours | RMSP | Report | Systems affected, with clear disruption and customer impact |
| Attacker uses an MSP's remote management tool to reach customer systems, before any outage | RMSP | Likely report | Security affected, and customers' systems are a listed factor |
| Phishing email blocked by the mail gateway | Any | Not an incident | No actual adverse effect |
| Stolen credentials used on one internal account, contained with no route to services | RMSP | Judgement call; record your reasoning | Security affected, but significant impact may not be likely |
| Attack on cooling controls is stopped before any effect on systems or service | Data centre | Not reportable under the Bill | No actual effect. Consider the future voluntary route |
| Attacker inside a hospital supplier's network with persistent access, no disruption yet | OES | Likely report | Security already compromised and significant impact likely |
What about near misses?
They are not mandatory to report under the Bill. At Report Stage, the minister committed instead to "a voluntary near-miss reporting mechanism for all regulated entities", to be set up without primary legislation. It doesn't exist yet.
Keeping an internal record is still worth doing. It shows your controls working and gives you something ready for the voluntary route. Our near-miss log template and breach vs incident vs near miss guide cover this.
When does the reporting clock start?
When you are first aware that a reportable incident has occurred or is occurring. Both the 24-hour initial and 72-hour full notifications run from that moment. The Bill doesn't define awareness further, or say how a regulator will treat time spent investigating. In our view, a defensible approach is to:
- Log when an event was first detected and by whom.
- Set a short internal target for deciding whether it is reportable, such as two hours.
- Record who decided, when and why.
- Treat a likely significant incident as reportable rather than waiting for certainty.
The incident reporting rules set out the full process. The 24-hour runbook and 72-hour report template put this into a timed process. If personal data is involved, check the separate UK GDPR test too. See incident reporting and UK GDPR.
How should you set your own thresholds?
Turn each factor into an escalation trigger for your own services, so the decision isn't made from scratch at 2am. For example:
| Factor | Example internal trigger |
|---|---|
| Disruption | Core service degraded for more than an agreed share of customers |
| Duration | Outage expected to exceed an agreed number of hours |
| Users' systems | Any confirmed attacker access to a customer environment |
| Data | Any confirmed loss of integrity or confidentiality of customer data |
| Society or economy | Any customer that is itself an OES affected |
These triggers prompt a decision. They aren't legal thresholds, so review them when regulators publish guidance under clause 19.
Sources
Frequently asked questions
No. It lists factors such as disruption, users affected, duration, geographical area, data compromise, impact on users' systems and impact on the economy or society. The Bill also revokes the retained EU regulation behind today's numeric thresholds for digital services. Regulators may add detail in guidance.
No. At Lords Report Stage the Government narrowed the incident definition to events with an actual adverse effect, and removed mandatory near-miss reporting for data centres. The minister committed to a voluntary near-miss reporting mechanism for all regulated organisations, set up without primary legislation. It doesn't exist yet.
When the organisation is first aware that a reportable incident has occurred or is occurring. Both the 24-hour initial notification and the 72-hour full notification run from that same moment. Log when an event was detected and when you decided it was reportable, with your reasons.
No. The two regimes have different tests. Data compromise is one factor under the Bill, while UK GDPR asks about risk to people's rights and freedoms. One event can trigger both, so check each test separately and send separate reports where needed.
Often, yes. Ransomware usually has an actual adverse effect on systems, and the minister cited it as a type of incident the Bill is meant to capture. Whether it is significant depends on the factors, such as disruption, users affected and duration.