Private digital intelligence bureau Lawful OSINT · Cyber monitoring · Due diligence
Illustration of an analyst comparing reports and statistical tables

Open data · 2025 edition

2025 public digital threat observatory.

Comparable facts where possible, explicit scopes everywhere. This page brings together selected official indicators without adding incompatible datasets or presenting them as IIS incident statistics.

Selected indicators

Three official viewpoints, three distinct scopes.

1 366incidents known to ANSSI in 2025
4 875incidents analysed by ENISA over its stated period
504,810assistance searches through Cybermalveillance.gouv.fr and 17Cyber in 2025 (all audiences)
60%of observed initial-access vectors attributed to phishing by ENISA

Reading rule: Scopes and definitions differ between organisations. These values must not be added together.

ANSSI · France

Events known to the agency: 2024 and 2025.

Selected ANSSI indicators for France, as published for 2024 and 2025
Metric20242025Interpretation
Security events4 3863 586Published total, not recalculated by adding the next two rows.
Reports3 0042 209Events that did not necessarily become confirmed incidents.
Incidents1 3611 366Confirmed handling category used by the publisher.

ANSSI recorded 460 possible data-leak events in 2025. It could confirm a leak in 42% of those events.

ANSSI reports that four sectors accounted for 76% of the 1,366 incidents known to the agency. The individually rounded shares were 34% for education and research, 24% for ministries and local authorities, 10% for health and 9% for telecommunications. The figures cover ANSSI’s reporting scope.

ANSSI 2025 report, page 7 (PDF in French)

ENISA · European Union

Attack volume does not equal operational impact.

DDoS share77%77%
Phishing in initial access60%60%
Vulnerability exploitation in initial access21.3%21.3%
NIS2 essential entities53.7%53.7%

ENISA analysed 4,875 incidents from 1 July 2024 to 30 June 2025. Approximately 80% were associated with hacktivist activity in the published analysis, while the agency notes that only 2% of hacktivist incidents caused service disruption. Percentages must therefore be read with both denominator and consequence in mind.

ENISA · 22 September 2026

A newer EU report covers a different observation window.

The ENISA Threat Landscape 2026, published on 22 September 2026, analyses events observed from 1 January to 31 December 2025. It supplements the 2025 edition above, whose window runs from July 2024 to June 2025.

51 %Recorded cases involving DDoS
32 %Public administration among targeted sectors
73 %Essential and important entities under NIS2

Selected published values. The period overlaps the 2025 edition; the two editions do not form a directly comparable series. The 2026 NIS2 categories combine essential and important entities.

Read the full ENISA 2026 report.

Cybermalveillance.gouv.fr · France

Assistance demand reveals the pressure felt by users.

≈ 33%of individual assistance requests related to phishing
15 000approximate individual assistance searches for fake bank advisers (+159%)
13 000approximate individual assistance searches for bank-transfer fraud (+196%)
+517%change in individual assistance searches for phone-number spoofing

Methodology

How this observatory is built and should be used.

IIS transcribed a limited set of headline indicators from the public publisher pages linked below. Values retain their publisher, period, wording and scope. No extrapolation was applied, and no incompatible count was added to another.

The tables distinguish headline values from sector distributions. Before comparing two indicators, check their unit, observation period, territory and counted population against the primary publication.

United Kingdom: from reports to incidents requiring support

From 1 September 2025 to 31 August 2026, the UK National Cyber Security Centre’s incident management team received 1,831 reports of cyberattacks affecting organisations. After triage, 337 incidents required support from the team, including 80 that it classified as nationally significant. These counts describe the work of that team over a twelve-month reporting period. Use the dates and reporting criteria before comparing them with figures for calendar year 2025.

NCSC Annual Review 2026, incident management (3 November 2026)

Data quality and interpretation limits

Published ANSSI reports and incidents add up to 11 fewer than the event total in 2025 and 21 fewer in 2024. Cybermalveillance counts assistance searches. The reason-specific figures shown apply to individuals.

The assistance count uses the 2025 report’s chart on page 31. Since the launch of 17Cyber, the publisher counts only the assistance journeys completed through its two online entry points.

Checks before comparing or adding figures

For ANSSI 2025, the source publishes 3,586 events, 2,209 reports and 1,366 incidents. The two latter values total 3,575. This observatory retains the published values and does not invent an explanation for the difference of 11. Do not reconstruct the total by addition.

The ENISA 2025 publication also carries a version 1.3 revision notice dated 22 September 2026. The indicators here are sourced to the linked publisher summary; for detailed analysis, consult the revised report and identify the version used. Report and revision notice.

Download the source-linked indicators

Both JSON files retain the publisher, observation period and interpretation notes. Keep these fields with the numbers when reusing the data.

Assess the urgency of an incident using its actual impact

FAQ

Read the figures without overstating them.

Can these figures be added together?

No. Each organisation uses a different population, reporting route, period and definition. The figures describe separate scopes.

Whose incident figures does the observatory use?

The observatory uses selected public indicators from ANSSI, ENISA, Cybermalveillance.gouv.fr and the UK NCSC. It does not include IIS mission data.

Why can a percentage and a count appear together?

A count describes volume in one dataset; a percentage describes a share or change within its own stated denominator. Read each metric with its scope and source.

Data literacy

A percentage needs a denominator

Before comparing two indicators, identify the counted unit, population, observation window and collection method. A shared publication year does not establish a shared period.

A share is not a probability

A share within observed incidents does not estimate the likelihood that your organisation will be attacked. The population includes recorded cases, rather than all organisations exposed to a threat.

A rise is not a share

In a fictional example, 100 requests rising to 150 represent a 50% increase. That does not mean 50% of all victims made that request: their total number has not been given.

A percentage point is a difference

In a fictional example, a share rising from 20% to 30% gains 10 percentage points. Its relative increase is 50%, provided both measurements use comparable definitions and populations.

Can 60% be multiplied by 4,875 incidents?

Not without confirming the denominator. Here, 4,875 describes the incidents analysed in the ENISA 2025 edition, whereas the phishing percentage concerns observed initial-access vectors. Applying the latter share to the former total would create a count that the source has not established.

Source version matters: the ENISA 2025 publication records a version 1.3 revision dated 22 September 2026. When citing a figure, retain the edition, version and observation period. ENISA Threat Landscape 2025.

Four people discussing how to organise their group — AI-generated photograph

From indicators to exposure

Public statistics provide context. A scoped assessment determines what is actually relevant to your organisation.

Laptop, phone and notebooks on a desk — AI-generated photograph
Scope an assessment