Cyberlutions
Back to homepage

Our approach

How we build the intelligence

Useful intelligence needs clear sources, honest uncertainty and figures that mean what they say. This is how to read our records and understand their limits.

Global coverage · Australian focusMethodology updated 6 October 2026

A record is an account of the available evidence.

Check its source, status, dates and qualifications before drawing conclusions. “Reported”, “suspected” and “confirmed” are not interchangeable, and a resolved incident does not mean every consequence has ended.

Start with the source

We bring together organisation disclosures, regulator and government publications, court material, credible reporting and breach catalogues. Original disclosures are preferred where available. Sources such as Have I Been Pwned and DeHashed provide catalogue metadata; their inclusion alone does not independently confirm an incident.

We look for supporting sources and distinguish the date of the incident from the date it was disclosed or added to a catalogue. If only one source is available, or a date is approximate, that limitation matters. Government statistics may describe aggregate trends rather than identify individual breaches.

Separate reporting from confirmation

Review considers the organisation, event, dates, reported impact, source attribution and possible duplicates. A published record can describe a reported or suspected incident; publication is not a guarantee that every claim has been independently verified.

Evidence confidence describes the supporting material and its relevance to the record. A high rating is not confirmation of every allegation, date, attribution or impact estimate. An authentic dataset can still contain uncertain provenance or overlapping accounts. Read the incident status and source qualifications alongside the rating.

Attacker claims and provider descriptions remain attributed. Unverified exposure listings, including Cit0day collection metadata, are kept in the separate Research catalogue. A filename or domain match is a research lead, not proof that the named organisation suffered a breach. Known fabricated datasets are excluded from published breach totals.

Keep different records separate

The incident catalogue describes published breach and incident records. The Intelligence Library holds related campaigns, advisories, enforcement actions and research. The Research catalogue holds source listings that can require further investigation. Library and Research entries do not add to the tracked breach total.

We reconcile reports against existing incidents before adding records, linking related material where appropriate. One organisation can have several distinct incidents, and one incident can have several sources. Duplicate reports should not become additional incidents; unresolved matches remain subject to review.

Understand the numbers

Tracked records
Published incident records, not the number of sources, archive files or unique organisations. Separate intelligence and exposure listings are excluded.
Reported people
Reported individual counts, including labelled estimates and claims. Unknown values are omitted. Totals are not unique people across incidents or an Australian-only count; accounts, files and documents are different measures.
Compromised accounts
Eligible HIBP-reported account counts across published records. Accounts can recur across datasets. This is not a count of unique people. Labelled unverified datasets may be included; fabricated and reviewed non-compromise datasets are excluded.
Reported data volume
Reported or claimed file sizes, including estimates and lower bounds. Known linked duplicate volumes are excluded. This is not an independently measured total.
Dates and ordering
Latest Incidents prioritises the incident date, then the disclosure date or known year where needed. Loading a record does not make the incident recent. Record years use the known incident start year; disclosure and catalogue dates remain separate. Where the source evidence conflicts, the incident date or year can remain unknown. Date precision depends on the source.

Preserve what was available

Archives may include a dated platform-record snapshot, a captured source page or an original document. These are different evidence types. A PDF made from a record or source extract is not the same as an original PDF issued by the source, and an archive does not establish that its contents are true.

Coverage varies. Some archives are retained locally, while others are stored privately online. Missing, unavailable and paywalled material can limit a capture to accessible text or metadata. A source link is not proof that the complete article or document has been archived. Master archives remain restricted; public links do not grant download access.

Describe coverage accurately

Coverage is worldwide, with an Australian focus. Australian relevance requires a documented basis, such as an Australian organisation or reported Australian impact. A global service or overseas incident is not automatically evidence of Australian victims.

Locations generally describe the organisation or incident, not every affected person's residence. State counts can overlap, and national incidents are not assigned to every state. Where a location, sector, attack method or impact cannot be established, it remains unknown or unclassified rather than inferred.

Leave room for correction

Reports develop over time. Dates, counts, classifications and source availability can change as further evidence becomes available. Our coverage reflects the material reviewed and is not a complete measure of all cyber incidents or their prevalence.

If a record is inaccurate, provide its link, the detail in question and supporting evidence through Factual Corrections. Use Takedown & Corrections for broader content-review concerns. Requests are reviewed; submitting one does not automatically change or remove a record.