Version 1.0 · Public Research Edition · TLP:CLEAR · Brandefense Country Report · Belgium 2026
About this editionAI-assisted research and drafting. Evidence labels and coverage limits apply throughout. The English edition prevails where translated editions differ.
Chapter 1
Executive summary
This chapter answers
What matters most for Belgian organisations in 2026, and what should leadership do first?
In short
Belgium's evidence points to identity protection and service continuity. The CCB received 588 incident reports through 28 September. Ransomware sites carried 46 Belgian listings within the report window. Retained forum posts advertised customer data, identity material, access and fraud tools. Public incidents affected care, membership information, education and administrative services. These populations measure different things. Validate active identities, map supplier dependencies, restrict remote operational access and rehearse notification and recovery decisions.
The report combines underground observations, exposure data and verified public reporting. Brandefense Blackhole records 96,265 country-tagged posts across platforms. The retained content covers 1,025 distinct forum posts, including 480 country-tagged forum posts. These overlapping populations describe visibility, not national attack counts. [S1][S2][S3]
1. Identity protection has the strongest official support. In its 9 November assessment, the CCB described account-related incidents as around 30% of reported incidents during the first two quarters. Police reporting separately describes reused mailbox credentials, invoice alteration and bank-impersonation fraud. Neither source measures current exposure at every Belgian organisation. [S5][S31][S32]
2. Extortion claims span many group labels. Ransompedia recorded 46 accepted listings dated within the window, across 22 group labels. Qilin accounts for 9 and The Gentlemen for 7. The partial-year total exceeds the 42 recorded during all of 2025. This is not an annual growth rate or proof of 46 distinct intrusions. [S1]
3. Shared suppliers turn technical incidents into service decisions. Learning, healthcare and local-government cases show how a dependency can affect communication, availability and data assessment. A customer population is not an affected-victim count. Require a supplier-specific and customer-specific impact statement. [S8][S13][S22]
4. Advertising mixes old data, current claims and untested capabilities. One telecom advertisement points back to July 2025. A large citizen-data poster admits its source is unknown. Recent posts advertise support-ticket data, bank impersonation tools and a casino-themed customer list. They do not establish working access or completed transactions. [S3]
5. Operational recovery needs its own owner. November refrigeration disruption creates a different problem from a website outage. Public attribution remained disputed in the retained reporting. The immediate defensive question is how to control remote access and return equipment to a safe state. [S25][S26][S27]
6. Current listings require scoped identity validation. 63 of 250 selected organisation domains had a credential listed for sale when checked. The national Atlas headline is 48 out of 100. At least 296,591 device associations and 26,236,674 credential records describe overlapping exposure populations, not confirmed intrusions. [S2]
Four decisions for leaders
- Validate identities before prescribing disruptive response. Ask identity and SOC leads to distinguish an old credential, an active account and a suspicious session. Keep a recorded disposition for each relevant finding.
- Map supplier dependencies to data and service impact. Ask each critical provider for scoped notification terms and a tested alternative process. Rehearse an outage while the confidentiality assessment remains uncertain.
- Control remote operational access. Give operations a verified access list, a safe local process and authority to restrict supplier access when needed. Test recovery without the ordinary remote route.
- Prepare clear notifications and recovery decisions. Assign privacy, legal, communications and technical owners. Keep statutory clocks separate from internal planning targets and preserve uncertainty across French and Dutch notices.
The year in numbers
Incident reports received by CCB, 1 January–28 September 2026 [S4]
What it is notA full-year total, unique victims or a national prevalence estimate
Accepted Belgian ransomware listings within the report window [S1]
What it is notConfirmed encryption, distinct intrusions or measured loss
Group labels attached to those listings [S1]
What it is notVerified separate criminal organisations
Distinct retained forum posts across the collected sets [S3]
What it is notAll posts concerning Belgium or unique incidents
Country-tagged forum posts within that retained population [S3]
What it is notAn additional population to add to 1,025
Retained posts dated 9–22 November [S3]
What it is notNot verified new intrusions
Purposively selected candidate domains [S2]
What it is notA ranked economic census or a representative national sample
Candidate domains with a DNS response at the observation date [S2]
What it is notValidated independent legal entities or a security rating
Selected organisation domains with a current credential listing [S2]
What it is notWorking access, active accounts or confirmed intrusions
National Atlas headline for the selected 250 domains [S2]
What it is notNational prevalence or a sector rating
Country-tagged posts across platforms [S3]
What it is notVerified Belgian targeting, criminal relevance or national incidents
Key takeaways
- 01
Prioritise account validation and recoverable services.
- 02
Keep advertisements, listings and confirmed incidents as separate populations.
- 03
Give suppliers, operations and notification decisions named owners.
Report map
Start with the decision you own, then follow the evidence and its limits.
Contents and reading guide
| Chapter | Question |
|---|---|
| 1 Executive summary | What should leaders decide first? |
| 2 Context | Which dependencies shape the Belgian picture? |
| 3 National exposure | What does the selected domain assessment establish? |
| 4 Dark web | What was advertised, and what remains unverified? |
| 5 Incidents | Which service and data outcomes were supported? |
| 6 Disruption | How do claims differ from operational impact? |
| 7 Actors | Which relationships are evidenced? |
| 8 Fraud | How could exposed information support deception? |
| 9 Banking | What should identity and fraud teams validate? |
| 10 Regulation | Which duties and milestones apply? |
| 11 Outlook | Which conditional scenarios deserve rehearsal? |
| 12 Action plan | Who owns the next 30/60/90 days? |
| Appendices A–D | Scope, evidence register, sources and glossary |
Leadership can start with Chapters 1, 5, 10 and 12. SOC and identity teams should read Chapters 3, 4, 7 and 9. Operations, procurement and customer-protection teams should use Chapters 5, 6 and 8.
Observed means a retained advertisement, listing or public statement exists. Claimed preserves what an advertiser says without validating its outcome. Assessed identifies a proportionate interpretation, its alternative explanation and the action it supports.
Chapter 2
Belgium's digital and threat context
This chapter answers
Which features of Belgium matter for incident response, and which conclusions should defenders avoid?
In short
Belgium's public services, commercial platforms and suppliers share digital dependencies across organisational and national borders. The CCB received 588 incident reports through 28 September. Its November identity report puts account protection at the centre of the response. The cases in this report also show why a supplier interruption can become a customer-service problem. Identify the account, supplier and operational dependencies that your organisation needs to keep working.
2.1 A digital dependency can cross several boundaries
Belgium's official portal describes a federal state with three communities and three regions. Dutch, French and German are official languages. These are administrative facts, not explanations of security performance. [S6]
For a defender, the useful question is who controls a service and its data. A central contract may support several local administrations. A foreign platform may hold Belgian customer records. An incident at that platform does not establish an intrusion into every Belgian customer's own network.
The report therefore separates domestic organisational incidents, shared-service incidents and foreign-service context. It does not rank language communities, regions or administrations by security. The selected domain observations cannot support such comparisons.
Communications should follow the affected users. Prepare plain-language notices in the languages your service uses. Keep the scope, uncertainty and protective advice identical across editions. Translation must not turn a possible disclosure into a confirmed theft.
2.2 Official reports point to identity and system exposure
The CCB received 588 incident reports between 1 January and 28 September 2026, against 556 during all of 2025. Those periods differ. The comparison shows that the earlier full-year total was exceeded; it is not an annual growth rate. [S4]
The quarterly reports describe changing incident mixes. The first quarter recorded 154 cyber incidents; the second recorded 188 cyber incidents within a wider notification population. These figures are not interchangeable with all notifications. [S7][S8]
The third-quarter report highlights system intrusions and exposed vulnerabilities. The CCB's November identity assessment says account-related incidents represented around 30% of reported incidents during the first two quarters. Reporting populations and periods remain attached to each figure. [S9][S5]
This supports two parallel workstreams. Identity teams should verify sessions, recovery routes and privileged access. Infrastructure teams should know which exposed systems require urgent remediation. Neither team can close its work solely by completing the other's checklist.
2.3 Suppliers create service dependencies
The CCB's second-quarter report discusses software supply chains affecting Belgium. Public cases concern healthcare software, learning platforms and a digital-signature service used by local authorities. Their common feature is dependency, not evidence of one coordinated attacker. [S8][S13][S22]
For suppliers, the first useful deliverable is a customer-specific impact statement. For customers, it is a record of which data, accounts and processes depend on that supplier. A supplier's client count is not the number of clients whose data was taken.
Keep commercial continuity separate from confidentiality. A service can be unavailable because it was deliberately isolated. Another service can remain available while personal data is exposed. Chapter 5 records these outcomes separately.
2.4 Politically framed activity needs careful attribution
The CCB describes state-sponsored activity and geopolitical hacktivism as relevant threats. That assessment does not attribute every Belgian incident to a state. The public record for the refrigeration campaign illustrates why the distinction matters. [S8][S9][S26]
Reuters reported different ministerial assessments on 13 November. The interior minister described Iranian involvement as very likely; the foreign minister said the available information was insufficient for official attribution. An investigation was continuing. [S26]
This report makes no independent attribution for that campaign. It takes no position on government decisions, international disputes, language politics or historical relationships. Actor rhetoric does not establish motive, technical responsibility or impact.
EU institutions appear only as public context. A Brussels location, an EU-related brand or a multilingual post does not by itself make an incident part of Belgium's domestic organisational population.
Key takeaways
- 01
Attach each official figure to its own reporting population and period.
- 02
Map supplier dependencies to the service and data actually affected.
- 03
Preserve uncertainty when describing attribution or translating a notice.
Chapter 3
National exposure: selected organisations and domains
This chapter answers
What do measured exposure signals mean for the selected organisations?
In short
63 of 250 selected organisation domains had a credential listed for sale when checked. The national Atlas headline is 48 out of 100. The portfolio records at least 296,591 device associations and 26,236,674 credential records. These overlapping observations do not establish unique people, active accounts or confirmed intrusions. Passive checks cover 245 of these domains. Validate current listings, investigate employee-associated devices and verify recovery routes.
3.1 Universe card
| Attribute | Scope |
|---|---|
| Population | 250 selected primary domains across ten sectors; 245 with passive observations |
| Unit | A primary domain; legal-entity independence and ownership are not established by DNS |
| Selection | Public-service and economic roles; a purposive portfolio, not a ranked national census |
| Exposure date | Cumulative device, credential, listing and document observations: 21 November 2026 |
| Other periods | Mentions: previous 365 days through 22 November 2026; lookalikes: 30 days; passive observations: 22 November 2026 |
| Overlap | Records can overlap across sources, domains, devices and people; totals are not unique-user counts |
| Interpretation | Findings describe selected domains, not national prevalence or every organisational asset |
The national scope of this report does not make the portfolio representative. A domain may serve employees, students, customers or several operating units. Domain associations do not establish device ownership or enterprise-network intrusion. [S2]
3.2 Exposure signals by sector
The national Atlas headline for the selected 250 domains is 48 on a 0–100 scale. Higher values describe more observed external exposure in this portfolio. The table reports signal counts by sector. [S2]
Cells show domains with the signal out of 25 selected domains per sector. Current credential listings are shown in coral.
Source: exposure observations for 250 selected organisation domains, 21 November 2026. Cumulative, overlapping associations; not confirmed intrusions or national prevalence [S2].
| Sector | Domains | Current listings | Listing history | With employee devices | With documents |
|---|---|---|---|---|---|
| Banking and finance | 25 | 11 | 11 | 18 | 18 |
| Government and public services | 25 | 3 | 3 | 11 | 11 |
| Local administration | 25 | 2 | 2 | 23 | 18 |
| Education and research | 25 | 6 | 6 | 21 | 19 |
| Health and care | 25 | 2 | 2 | 19 | 13 |
| Energy and utilities | 25 | 2 | 2 | 16 | 9 |
| Transport and logistics | 25 | 8 | 8 | 18 | 13 |
| Industry and business | 25 | 2 | 2 | 20 | 20 |
| Retail and consumer services | 25 | 18 | 18 | 16 | 14 |
| Telecom, technology and media | 25 | 9 | 9 | 17 | 18 |
| All selected domains | 250 | 63 | 63 | 179 | 153 |
The last two columns count domains with an employee-device association or exposed-document reference, not devices or documents. Every sector contains 25 selected domains. A current listing supports account validation; it does not establish that an account works. [S2]
3.3 Devices and credentials
- Employee-device associations
- Other device associations
Device associations (at least)
Credential records (overlapping)
Source: exposure observations for 250 selected organisation domains, 21 November 2026. Cumulative, overlapping associations; not confirmed intrusions or national prevalence [S2].
| Sector | Domains | Devices, at least | Employee devices | Credential records |
|---|---|---|---|---|
| Banking and finance | 25 | 13,221 | 156 | 190,956 |
| Government and public services | 25 | 11,202 | 53 | 44,270 |
| Local administration | 25 | 4,119 | 336 | 137,471 |
| Education and research | 25 | 21,825 | 2,833 | 652,567 |
| Health and care | 25 | 8,484 | 138 | 85,104 |
| Energy and utilities | 25 | 25,718 | 93 | 32,333 |
| Transport and logistics | 25 | 46,548 | 1,849 | 658,184 |
| Industry and business | 25 | 4,545 | 1,294 | 404,814 |
| Retail and consumer services | 25 | 65,121 | 118 | 41,710 |
| Telecom, technology and media | 25 | 95,808 | 14,023 | 23,989,265 |
| All selected domains | 250 | 296,591 | 20,893 | 26,236,674 |
The portfolio contains at least 296,591 device associations, including 20,893 classified as employee-related. Credential records comprise 9,869,372 breach-database records, 1,488,704 stealer-log records and 14,878,598 combolist records. These sources overlap; their total does not count unique accounts. [S2]
Some device totals are lower bounds. An employee classification is an investigative lead, not verified employment or device ownership. Shared educational and customer-facing domains need role validation before account response.
The portfolio also records 38,774 lookalike-domain candidates and 1,333 exposed-document references. Similarity does not establish malicious use; a document reference does not prove confidential data was disclosed. Dark-chat mentions occur for 39 domains and dark-web mentions for 31. Visibility does not establish hostility. [S2]
3.4 How to handle a current identity finding
Assign a finding to an accountable identity owner. Check whether the account exists, remains active and has already been remediated. Compare the finding's age with recent authentication and recovery activity.
If a suspicious current session is verified, revoke it and review persistence, registered authentication methods and mailbox changes. If the record is historical and inactive, document that disposition. Preserve evidence before deleting or resetting an account.
The 63 current credential listings concern the selected domain portfolio. Forum requests and capability advertisements are separate observations. Neither a current listing nor a buyer request establishes working access or a confirmed organisational intrusion. [S2][S3]
3.5 Email and web controls
| Observation | Count | Denominator |
|---|---|---|
| DMARC reject or quarantine | 188 | 245 |
| DMARC monitoring policy | 46 | 245 |
| No valid DMARC policy observed | 11 | 245 |
| SPF observed | 234 | 245 |
| DKIM information observed | 204 | 245 |
| DS (delegation signer) record observed | 65 | 245 |
| TLS certificate validated | 240 | 245 |
| TLS certificate error | 1 | 245 |
| TLS outcome undetermined | 4 | 245 |
| Homepage response obtained | 213 | 245 |
| HSTS (HTTP Strict Transport Security) in obtained response | 164 | 213 |
| CSP (Content Security Policy) in obtained response | 91 | 213 |
A published DMARC policy describes how receivers are asked to treat unauthenticated messages. A monitoring policy differs from enforcement. Even an enforcement policy is not proof that every legitimate sender is aligned or that every receiver applies it. [S2]
SPF and DKIM support email authentication, but their presence is not a complete mail-security assessment. Failure to observe a signing configuration does not prove that no valid signing configuration exists. Owners should validate their actual sending services.
A DS record observation is not end-to-end validation of DNSSEC. A successful TLS observation concerns the assessed host at that time. A failed or missing observation must be reviewed before it is described as an organisational configuration defect. [S2]
Homepage headers describe the captured response, rather than every application or path. An absent header can justify a configuration review; it cannot establish exploitability, sensitive-data exposure or a successful intrusion.
Source: passive observations, 22 November 2026. Each cell shows observed/assessed domains; published policies do not establish enforcement for every message [S2].
| Sector | Observed | DMARC reject/quarantine | SPF observed | TLS valid |
|---|---|---|---|---|
| Banking and finance | 25 | 23 | 25 | 25 |
| Government and public services | 25 | 21 | 20 | 24 |
| Local administration | 23 | 20 | 22 | 22 |
| Education and research | 24 | 13 | 23 | 24 |
| Health and care | 24 | 16 | 21 | 23 |
| Energy and utilities | 25 | 21 | 25 | 25 |
| Transport and logistics | 24 | 16 | 23 | 24 |
| Industry and business | 25 | 22 | 25 | 23 |
| Retail and consumer services | 25 | 19 | 25 | 25 |
| Telecom, technology and media | 25 | 17 | 25 | 25 |
Key takeaways
- 01
Validate current listings and employee-device associations with their owners.
- 02
Keep overlapping records separate from people and confirmed intrusions.
- 03
Review passive controls against the assessed domain and date.
Chapter 4
What we see on the dark web
This chapter answers
What do retained advertisements establish about Belgian data, access and fraud demand?
In short
The retained forums show a market for customer records, identity images, access and fraud tools. Some advertisements explicitly refer to older incidents. Others acknowledge uncertain provenance or describe proposed data structures. The closing weeks add support-ticket, bank-template and casino-themed claims. The observed fact is the advertisement. Validate relevance, freshness and access through trusted organisational evidence before treating an offer as a current incident.
4.1 Universe card
| Attribute | Scope |
|---|---|
| Period | Post dates 1 January–22 November 2026 |
| Retained population | 1,025 distinct posts in the retained forum material |
| Country-tagged subset | 480 posts, included within the retained population |
| Recent subset | 31 posts dated 9–22 November, also included within the retained population |
| Venue | Retained content is forum-based; tagged counts also cover other platforms |
| Unit | A post, not a unique file, seller, purchase, victim or incident |
| Coverage limit | Some advertised records remain unavailable; tagged totals and retained forum posts are separate populations |
The forum and country-tagged populations overlap. Their totals must not be added together. Country labels and references to Belgium do not independently validate targeting, provenance or nationality of a victim. [S3]
Brandefense Blackhole records 96,265 country-tagged posts across platforms during the report window. A country tag does not establish Belgian targeting, criminal relevance or an incident. These counts include discussion and advertising; they are separate from the 1,025 retained forum posts. [S3]
- Country-tagged posts
Source: Brandefense Blackhole, 1 January-22 November 2026; November partial. Tagged posts and retained forum posts are separate populations, not confirmed incidents [S3].
| Month | Country-tagged posts |
|---|---|
| 2026-01 | 2,299 |
| 2026-02 | 2,322 |
| 2026-03 | 2,659 |
| 2026-04 | 3,405 |
| 2026-05 | 1,610 |
| 2026-06 | 2,141 |
| 2026-07 | 3,356 |
| 2026-08 | 44,206 |
| 2026-09 | 18,132 |
| 2026-10 | 14,312 |
| 2026-11 | 1,823 |
- Country-tagged posts
Source: Brandefense Blackhole, 1 January-22 November 2026; November partial. Tagged posts and retained forum posts are separate populations, not confirmed incidents [S3].
| Platform | Country-tagged posts |
|---|---|
| telegram | 95,744 |
| forum | 480 |
| discord | 17 |
| paste | 24 |
| tor | 0 |
Telegram accounts for 95,744 of the 96,265 tagged posts. Retained case evidence comes from forums; the messaging population is represented here by counts. Those counts do not validate advertised claims or measure attack frequency. The forum content cannot establish crime-family shares across all platforms. [S3]
4.2 Volume and language
| Month | Retained posts |
|---|---|
| 2026-01 | 342 |
| 2026-02 | 92 |
| 2026-03 | 78 |
| 2026-04 | 94 |
| 2026-05 | 43 |
| 2026-06 | 66 |
| 2026-07 | 49 |
| 2026-08 | 50 |
| 2026-09 | 75 |
| 2026-10 | 56 |
| 2026-11 | 80 |
- Retained forum posts
Source: Brandefense Blackhole, 1 January-22 November 2026; November partial. Tagged posts and retained forum posts are separate populations, not confirmed incidents [S3].
The monthly figures describe this retained population. They do not establish that Belgian threat activity rose or fell with the number of posts. Advertising, reposts, discussion and visibility can each affect the count. [S3]
Research covered English, French, Dutch and Russian reporting, with additional relevant regional context. The language of an advertisement does not establish the advertiser's nationality or the language of affected users.
4.3 The products advertised
| Family | What the retained material shows | What it does not establish |
|---|---|---|
| Customer and citizen data | Telecom, training, legal-service and consumer-themed offers | Authenticity, unique people or fresh intrusion |
| Identity documents | Requests for Belgian identity images | A supplied document or successful onboarding |
| Access and sessions | Claims of administrative access and demand for cookies | Valid current access or a completed purchase |
| Bank-account identifiers | Advertised identifiers and asking prices | Account control or payment authority |
| Fraud tools | Bank-specific phishing-template advertisements | A working tool, deployed campaign or successful authentication interception |
These are observed themes, not mutually exclusive incident categories or shares of a national underground economy. One advertisement may mention several products. A transaction is not established merely because a price is displayed. [S3]
4.4 Case studies
Case A
A telecom advertisement points back to 2025
- Claim
- The advertiser offers a customer-data sample described as exceeding 300k people. It explicitly dates the underlying incident to July 2025.
- Chronology
- The retained advertisement is dated 24 January 2026. Similar titles and later reuploads do not independently establish identical files.
- Established
- The advertisement and its historical reference are retained. The record count, source and authenticity are unverified.
- Alternative
- Older material may be copied, relabelled or resold. A 2026 posting date does not establish a fresh intrusion.
- Response
- Validate affected accounts and prior remediation before describing a new incident. Prepare proportionate warnings if old personal details are circulating again. [S3]
- Held
- A January advertisement states an older incident date and advertises a customer-data sample.
- Not established
- The number or authenticity of records, a new 2026 intrusion, or completed transactions.
Case B
A citizen-data poster admits the source is unknown
- Claim
- A persona advertises 3.1M Belgian citizen records and suggests a public-service origin while admitting that the source is unknown.
- Chronology
- The 21 January post drew a reply suggesting older social-network data. The original persona accepted that as a possibility.
- Established
- The uncertainty is visible in the discussion. Neither the original claim nor the reply validates the data source.
- Alternative
- Historical social-media material, public-source collection and relabelled data remain plausible. The discussion does not establish that a government database was accessed.
- Response
- Require validated provenance and relevant samples before notifying a supposed source organisation. Do not use a large country-labelled count as the verification. [S3]
- Held
- A forum post claims a large population dataset while acknowledging an unknown source.
- Not established
- Government provenance, the size or uniqueness of a dataset, or a 2026 state-system intrusion.
Case C
An autumn training-network offer may recirculate older material
- Claim
- The advertiser offers more than 700K people and +85 K IBAN, while also claiming to sell access. Those are different propositions.
- Chronology
- The retained offer is dated 2 October. February replies and another October package refer to the same target topic. Public reporting separately describes republication. [S20]
- Established
- The advertisement is retained. Neither the number of distinct people nor current access is independently established.
- Alternative
- An older package can be reposted or advertised by a different persona. Similar naming does not establish file identity or a separate October intrusion.
- Response
- Validate present account access separately from historical data exposure. Warn affected users about renewed impersonation when the organisation has established relevance. [S3]
- Held
- An October advertisement makes both dataset and access claims. Earlier February replies reference the same target topic.
- Not established
- The number of distinct people, valid current access, or a separate October intrusion.
Case D
An access offer repeatedly describes proposed tables
- Claim
- A seller claims administrative access to an identity-verification service, approximately 412k records and a price of USD 4,550.
- Chronology
- The offer is dated 8 November. Its descriptions repeatedly call the contact, identity-check and recovery tables proposed.
- Established
- The schema descriptions and access assertion are retained. They do not show a validated account, tenant or database.
- Alternative
- The structures could be hypothetical, copied or synthetic. Their wording raises questions without independently proving fabrication.
- Response
- Use trusted provider channels to validate a relevant tenant. Preserve administrative activity and recovery records before prescribing broad account resets. [S3]
- Held
- The offer and its schema descriptions; the title claims roughly 412k records.
- Not established
- Valid access, actual stored records, the count, a successful sale or any supplier or identity-service incident.
Case E
A phishing-panel advertisement names bank templates
- Claim
- The SALTA3 advertisement markets templates for specific banks, including Belgian brands, and claims interactive authentication features.
- Chronology
- The 20 November post falls within the closing two weeks. It is tool marketing, rather than a retained attack trace.
- Established
- The advertisement and claimed brand coverage exist. A bank brand in a template list is an impersonation theme, not evidence of bank-system access.
- Alternative
- Features can be exaggerated or copied. The material does not establish working capability, deployment or successful authentication interception.
- Response
- Rehearse customer reporting and unusual device-enrolment or payment-approval handling. Review phishing-resistant authentication while preserving a clear distinction between marketing and observed attacks. [S3]
- Held
- A 20 November advertisement markets bank-specific templates, including Belgian brands, and claims interactive features.
- Not established
- A working kit, successful MFA interception, active deployment, victims, bank-system access or payment losses.
Case F
A supplier support-system claim includes tickets and attachments
- Claim
- A persona claims to hold all support tickets and attachments and advertises 6,028 users. Its claims about customers and earlier contact remain unverified.
- Chronology
- The post is dated 13 November. No retained evidence independently establishes the claimed earlier contact with the supplier.
- Established
- The assertion is retained. It does not establish possession of every attachment or the impact on a particular customer.
- Alternative
- The material could be partial, historical, copied or fabricated. A customer name inside advertising does not validate customer impact.
- Response
- Request a customer-specific supplier statement. Limit sensitive support attachments and review administrative access to portals that may hold operational details. [S3]
- Held
- An actor claims support tickets and attachments; the post advertises 6028 users.
- Not established
- That all tickets or attachments are held, affected customers, current access or technical impact.
4.5 Additional evidence cards
The remaining cards show identifier advertising, identity-image demand, session buying and a recent casino-themed offer. The original excerpts are retained advertisement text with identifying material removed. They are not instructions or validated product descriptions. [S3]
- Held
- A seller advertises bank-account identifiers with a country label and asking prices.
- Not established
- Access to accounts, usable payment authorisation, a named bank incident, or a completed sale.
- Held
- A buying request for Belgian identity documents.
- Not established
- Any supplied document, completed purchase, successful identity verification or source breach.
- Held
- A buying request names logistics platforms and active email cookies, with an asking range.
- Not established
- A seller response, purchased sessions, account validity, a completed sale or a platform breach.
- Held
- The 21 November advertisement and a displayed contact-data sample with claimed deposit fields.
- Not established
- The source casino, dataset freshness, deposit accuracy, current account access or a new intrusion.
An IBAN offer quotes USD 1.75 for 1,000 identifiers and USD 70.19 for the whole advertised package. The quantities are not an exact verified inventory. An IBAN is not authority to move money. [S3]
A logistics-session request displays a USD 250–2,500 range. It establishes demand and an asking range, rather than account validity or completed trade. The defensive relevance is session recovery and independent verification of payment changes. [S3]
4.6 Recent activity in the retained forums
| Post date | Card | Advertisement | Boundary |
|---|---|---|---|
| 2026-11-13 | B06 | Supplier support tickets and attachments | Customer names in a claim do not establish customer impact |
| 2026-11-20 | B05 | Bank-specific phishing templates | Marketing is not proof of deployment or a bank-system incident |
| 2026-11-21 | B10 | Casino-themed customer data | Similar September advertising does not establish identical files or a fresh incident |
The identity-service offer in B04 is dated 8 November and falls outside the recent period. Its proximity to later notices does not link it to them. No shared operator or common incident is inferred. [S3]
A 22 November advertisement again offered bank-impersonation templates and claimed interactive authentication features. Another multinational commodity advertisement included Belgium in its price list. These observations do not establish working access, verified card validity or a new bank intrusion. [S3]
4.7 What the forums add
The forums make the wording and timing of claims visible. A population-data poster concedes uncertainty about provenance. A telecom advertiser supplies an older incident date. An access seller repeatedly describes its tables as proposed.
Those details help an investigator ask better questions. They do not settle authenticity. A title, row count or schema can be copied, partially true, outdated or fabricated. The report preserves those alternatives instead of treating all advertisements as equivalent breaches.
Key takeaways
- 01
Post dates identify advertising, not intrusion dates.
- 02
Validate provenance and working access separately.
- 03
Recent fraud-tool marketing supports preparation, not a claim of successful attacks.
Chapter 5
Confirmed incidents and claims by sector
This chapter answers
Which outcomes were established, and which numbers still belong to a claimant?
In short
Belgium's public record covers interrupted healthcare, exposed membership information, education platforms, municipal systems and shared suppliers. Several organisations confirmed an incident while leaving the number of affected people unresolved. Restoration and data-loss assessments sometimes changed on different dates. The case cards preserve those distinctions. Use the closest operational example to test your own containment, customer notice and recovery decisions.
5.1 How to read the incident register
A confirmed incident does not confirm every criminal claim attached to it. The register separates direct organisational statements, official warnings, attributed reporting and underground advertisements. The cases below are selected dossiers, not a national census.
Some foreign services hold Belgian users' data. They belong in the dependency picture, but a worldwide user count is not a Belgian affected count. Repeated advertisements and related federation notices are not automatically independent incidents.
Names appear only when official confirmation and wider reporting support them. Other organisations have broad role labels. A role label protects identity; it must not conceal uncertainty about the event itself.
5.2 Healthcare and learning services
Case G
A hospital isolates systems in January
- Event
- A Belgian hospital network detected a serious IT disruption on 13 January. Its statement, quoted in reporting, described a precautionary shutdown across its campuses. [S10]
- Chronology
- Scheduled care was interrupted and critical patients were transferred. The retained January account does not establish a ransom payment, a named attacker or the full data-loss scope.
- Established
- Clinical continuity depended on the ability to work with digital systems unavailable. A shutdown decision can protect systems while creating immediate operational costs.
- Alternatives and limits
- Service disruption alone cannot distinguish encryption, containment or another technical cause. Later criminal claims require their own evidence.
- Response
- Clinical leadership should rehearse manual documentation, patient transfers and communication with neighbouring providers. Record which systems must be restored before normal scheduling resumes.
Case H
Basic-Fit confirms a cross-border customer-data incident
- Event
- Safeonweb relayed company confirmation on 13 April that Belgian members were among those involved in a customer-data incident. Reuters also reported the company statement. [S11][S12]
- Chronology
- The disclosure described information linked to members across several countries. The figure of about one million relates to the wider international population, not Belgium alone.
- Established
- Potentially affected data included contact, birth-date and bank information. The company said passwords and identity documents were not taken. [S11]
- Alternatives and limits
- A bank-account identifier is not evidence of access to the holder's bank account. The retained sources do not establish a Belgian affected-person count.
- Response
- Warn customers about messages that quote real membership details. Support teams should verify identity through a separate trusted process before changing contact or payment information.
Case I
A learning-platform incident raises Belgian dependency questions
- Event
- Safeonweb's 6 May warning described a vendor-confirmed incident at an online learning platform used in Belgium. The vendor's confirmation was dated 1 May. [S13]
- Claim
- A criminal group advertised a worldwide population of 275 million users. Safeonweb explicitly described that number as unconfirmed. It is not a Belgian count.
- Established
- Possible data categories included names, email addresses, student identifiers and messages. The vendor reported no indication at that time of password, birth-date, official-ID or financial-data involvement.
- Alternatives and limits
- A platform-wide claim cannot establish which records were taken from a particular institution. Institutional notifications and service restoration answer different questions.
- Response
- Education providers should request a scoped impact statement, prepare student notices and review staff support workflows. Treat unexpected course-related messages as a customer-protection concern.
The CCB's second-quarter report also discusses a healthcare-software supplier incident relevant to Belgium. That establishes a dependency concern; it does not establish that every client hospital suffered data theft. [S8]
5.3 Municipal and shared public services
Case J
Machelen's data-loss assessment changes after the initial incident
- Event
- The municipality identified a cyber incident on 11 September. Its own update says that further analysis confirmed personal-data theft on 22 September. [S15][S16]
- Chronology
- Digital services were affected. The captured municipal notice says service had become operational again, with possible delays, while the data-loss investigation continued.
- Established
- The incident date, confirmation date and restoration state are different facts. The initial uncertainty did not prevent the municipality from providing later updates.
- Claim
- A later extortion listing is recorded in the report's ransomware population. The listing does not independently establish the complete data set, attack method or identity of the intruder. [S1]
- Response
- Maintain an update page with dated changes. Give residents practical protective advice while explaining which data categories and affected people remain under investigation.
Case L
A digital-signature supplier affects local-government workflows
- Event
- Reporting on 26 October quoted a supplier's chief executive describing unauthorised access to a digital-signature service. Suspicious activity had been detected on 23 October. [S22]
- Chronology
- The supplier took the service offline and restored it the following evening with restrictions. Customer-specific impact assessments were being prepared.
- Established
- The supplier said mainly metadata was involved and said fraudulent signatures had not been created. Those are attributed company statements, not an independent forensic conclusion.
- Alternatives and limits
- More than 120 local authorities used the module. That customer population is not an affected-authority count. A separate integration population must not be added to it.
- Response
- Ask for the impact on your own documents, identities and integrations. Review replaced access permissions and test payment notices through a known official channel.
Case M
Federal tax services restore applications in stages
- Event
- SPF Finances/FOD Financiën confirmed a cyberattack on 28 October. Its recovery notice says it contained the incident within hours and temporarily took applications offline. [S23][S24]
- Chronology
- Main services returned on 31 October. A withholding-tax function returned on 6 November, and e604 on 12 November. These are restoration milestones, not separate intrusions.
- Established
- The authority stated that its investigation found no data exfiltration and no attacker control of applications. Attribute that conclusion to the authority and its investigation. [S23]
- Alternatives and limits
- A preventive shutdown is a real availability impact. It does not prove data theft, and restoration alone would not establish its absence.
- Response
- Exercise staged recovery and communicate which functions remain unavailable. Keep technical containment, user impact and confidentiality findings in separate sections of the incident record.
5.4 Membership, customer and operational data
Case K
A fitness-chain incident and a larger criminal headcount
- Event
- A fitness-club chain acknowledged an incident affecting membership information. Reporting quoted its spokesperson describing records linked to at most 31 of its 85 Belgian clubs. [S18]
- Claim
- The widely repeated figure of 150,000 members was attributed to the people advertising the data. It must not become a company-confirmed affected count.
- Established
- The reported categories included identity, contact and membership details, and some bank-account identifiers. The company said passwords and health data were not involved. [S18]
- Alternatives and limits
- Clubs, members, prospects and advertised records are different units. Dividing a claimed headcount by the club count would not establish exposure per club.
- Response
- Notify the people identified by the company's investigation and prepare for tailored impersonation. Do not authenticate a caller merely because they know a membership or bank-account detail.
Case N
Refrigeration disruption makes remote access an operational issue
- Event
- Official warnings and public reporting describe a November incident affecting retail refrigeration. Changed operating conditions created consequences for stores and stock. [S25][S27]
- Chronology
- Belgian disruption was reported on 2 November. Further technical advice and attributed statements followed. Different reports counted stores, equipment and suppliers' customer populations.
- Established
- Operations teams need to control remote access and retain the ability to restore safe settings. The public sources do not justify adding all reported equipment and store counts together.
- Attribution
- Reuters reported conflicting ministerial assessments on 13 November. This report does not resolve that disagreement or independently attribute the campaign. [S26]
- Response
- Immediately remove direct public internet access to refrigeration management interfaces. Operations and security should coordinate safe continuity and protected remote maintenance, restrict accounts to approved needs, and rehearse a safe local mode. Verify safe settings and retain evidence that the interfaces are no longer directly reachable. [S25]
Case O
A recent gift-voucher service notice
- Event
- Safeonweb warned on 20 November about unauthorised access to a gift-voucher platform. Its notice relayed the company's finding and said the investigation was continuing. [S29]
- Established
- Names and email addresses may have been viewed. According to the company, passwords and payment-card data were not affected at that stage.
- Alternatives and limits
- The notice did not provide a Belgian affected-customer count. The ability to send convincing follow-up messages is a plausible risk, not proof that a particular scam already used these records.
- Response
- Customer teams should publish a short warning on the service's normal channels. Advise users to open the official app or website directly when checking an unexpected message.
5.5 Other recorded incidents and reported claims
These rows preserve additional incident coverage without treating every account in the press as a verified technical finding.
| Period | Broad role | Evidence and limit | Defender implication |
|---|---|---|---|
| June | A sensitive public agency | Reporting described a mobile-management incident spanning 2025 and spring 2026; official confirmation and attribution were unresolved [S14] | Investigate management-plane access without assuming the internal intelligence network was reached |
| September | A sports federation | Reporting quoted confirmation of a membership-database incident; the complete scope was not independently established [S17] | Prepare membership notices and review outsourced administration |
| September | A related sports membership administration | A federation and its wing were discussed together; advertised record counts remain claims [S17] | Keep one campaign dossier when the evidence concerns related systems |
| October | A school group | A director confirmed an incident affecting pupils and staff; 3,550 plus 650 is 4,200 people in the reported population [S19] | Review staff access and support processes, including home-address sensitivity |
| October | A vocational-training network | Reporting described republication of data attributed to a February incident; 700,000 records and 85,000 bank details were reported figures [S20] | A repost can renew phishing risk without establishing a new intrusion |
| October | A public e-learning platform | Reporting described a service closure and a separate advertised data count [S21] | Treat availability and the data claim as separate validation tasks |
| November | An employee-benefits payment service | Official warning described possible account-data exposure; the company had not identified payment-system problems or suspicious transactions [S28] | Warn account holders without describing unverified payment loss |
| April | An online booking service | Safeonweb relayed possible reservation-data access and changed reservation PINs; the company said financial data was not accessed [S11] | No Belgian affected-user total was established |
5.6 Additional business, research and local-service cases
Case R
A research institute confirms copied data while services continue
- Official account
- A specialist medical and research institute's notice describes unauthorised access to and copying from a database after detection on 12 October. The notice was updated on 10 November. [S49]
- Established
- The institute says most copied information was coded, with a smaller amount directly identifiable. Its current findings did not identify effects on other systems; care and teaching continued.
- Limits
- Coded data is not automatically anonymous. A separately reported notification headcount does not measure the entire copied database, and the report supplies no combined affected-person total.
- Response
- Research and privacy teams should assess coding and identification risk together. Identify who holds the linking information, document the affected projects and prepare proportionate participant communication.
Case S
A professional body's contact population needs a precise denominator
- Official account
- A professional legal body's 6 November update says it was informed of the incident on 29 October. It describes access involving an administrative application and advertised contact information. [S50]
- Established
- The update identifies approximately 27,500 contacts who are not lawyers, in addition to lawyer records. It does not give one total affected population.
- Limits
- The body says its analysis did not identify banking or special-category personal data in the affected population. This is the body's assessment, not an independent verification of every advertised file.
- Response
- Keep professional and external-contact populations separate. Warn recipients about messages using genuine contact details and validate the scope before reusing an advertiser's larger count.
The following additional cases widen the service picture. They do not all establish data theft or ransomware.
| Period | Broad role | What the source supports | Limit |
|---|---|---|---|
| April | A municipality | The mayor described disruption and data theft; restoration was gradual [S53] | Original municipal statement not retained; no complete affected-person count |
| May | A chemical manufacturer | Its own clarification describes unauthorised access and says forensic work found no evidence of exfiltration or ransomware [S47] | Later publicity concerns the same May event |
| May | A recycling-park operator | Its follow-up scopes the incident to an external supplier computer; operations were isolated as a precaution [S48] | No indications of exfiltration in the stated investigation |
| July | A supermarket's external customer-data supplier | Reporting quotes a company statement about contact data on a separate supplier server [S51] | Do not generalise to payment systems or every customer |
| August | A furniture retailer | A spokesperson described short disruption and access to customer contact and order information [S52] | Criminal attribution remains a claim |
| October | A travel operator | A spokesperson confirmed a data incident and customer warning [S56] | The origin of personalised fraud messages remained unresolved |
| October, reported in November | A precious-metals dealer | Reporting quotes confirmation of customer-system access [S54] | A proposed connection to burglaries was not established |
| November | A municipal public website | A mayor described website access and possible staff-account exposure [S55] | No finding of citizen-data exposure or access to separate municipal services |
A charging-account service supplies recent cross-border context. Its own FAQ describes an 8 November attack, discovery on 10 November and remediation on 12 November. It notified users on 18 November. A Belgian affected count is not established. [S46]
VRT reports more than 3.5 million European users; that population is not an affected-account count. [S71] The company identifies account and profile information, including password hashes, while excluding several other categories in its notice. Individual Belgian service providers should validate their own dependency and users. [S46]
5.7 What these cases establish
The strongest repeated pathway is a gap between the service boundary and the organisational boundary. Members use a platform, staff rely on a supplier and residents depend on an administrative application. Incident response must identify the actual dependency.
The cases also show why clear reporting matters. A confirmation can narrow a criminal claim without eliminating every risk. A restoration notice can help users resume work while the data investigation continues.
The report does not add customer counts, claimed records, organisations and store counts into one headline. Such a total would measure nothing consistent. Leaders should ask which service, data category and person requires action now.
Key takeaways
- 01
Keep restoration, data exposure and attribution as separate findings.
- 02
Supplier client populations are not affected-victim totals.
- 03
Reposted data can renew fraud risk without proving a new intrusion.
Chapter 6
Hacktivism, disruption and operational systems
This chapter answers
What was observed, what was claimed, and how should services remain usable during disruption?
In short
The public record supports preparation for both website disruption and operational consequences. The CCB reported no hacktivist DDoS incidents in the first quarter and a limited return in the second. November refrigeration incidents show a different continuity problem. Actor statements and political explanations do not establish technical impact. Prepare status communications, service alternatives and controlled remote access before the next disruption.
6.1 Timeline
- 13 Jan 2026Hospital statement, as quoted in reporting
Precautionary system shutdown and interrupted care [S10]
- 22 Apr 2026CCB quarterly report
No hacktivist DDoS incidents reported for the first quarter [S7]
- 15 Jul 2026CCB quarterly report
A limited, one-day return of DDoS activity described for the second quarter [S8]
- 11 Sep 2026Municipal incident notice
Digital service interruption, followed by a later data-theft confirmation [S15]
- 28 Oct 2026Federal tax administration
Applications temporarily taken offline after an incident [S23]
- 2 Nov 2026
- 9 Nov 2026CCB identity assessment
Account protection remains a priority [S5]
- 10 Nov 2026Medical research institute
Its incident statement is updated; this is not a new intrusion [S49]
- 12 Nov 2026Federal tax administration
e604 restoration update [S23]
- 13 Nov 2026Reuters
Differing ministerial assessments of refrigeration attribution [S26]
- 17 Nov 2026Safeonweb
Warning about fake pellet and firewood offers [S45]
- 18 Nov 2026Charging-account service
Users notified after earlier detection and remediation; Belgian affected count unavailable [S46]
- 20 Nov 2026Safeonweb
Gift-voucher service warning relays ongoing investigation [S29]
An official row identifies the source of the account. It does not mean that the source independently verified every attacker statement mentioned in that account. A relay row identifies reporting rather than an original actor post.
6.2 Claim against public account
| Point | Claim or interpretation | Official account | Assessment |
|---|---|---|---|
| Website availability | An actor may advertise a site as unavailable | CCB reports describe their own observed incident population [S7][S8] | A screenshot or a post does not establish outage duration or national impact |
| Tax-system shutdown | Unavailability may be described as successful access | The authority says applications were isolated and no exfiltration or application control was found [S23] | Attribute the findings to its investigation |
| Refrigeration responsibility | A state may be named in political commentary | Reuters records differing ministerial assessments and an ongoing investigation [S26] | Attribution remains unresolved in this report |
| Refrigeration scale | Equipment, stores and customers appear in different accounts | Official warnings describe the operational risk [S25][S27] | Do not combine incompatible denominators |
6.3 A website incident is not an equipment incident
A website can be unavailable while internal operations continue. An operational interface can change equipment behaviour while the corporate website remains online. Their telemetry, owners and recovery decisions differ.
For public websites, agree how traffic mitigation is requested and how legitimate users retain access. Preserve availability measurements and the time zone. Avoid statements that infer data access from a denial-of-service claim.
For refrigeration equipment, immediately remove direct public internet access to management interfaces. Operations and security should coordinate safe continuity and protected remote maintenance. Verify that management interfaces are no longer directly reachable. [S25]
For operational equipment, involve the team responsible for safe settings. Test what happens if the supplier's remote route is removed. Verify that the local recovery process remains safe and that emergency access is logged.
The refrigeration case does not establish that all connected household appliances face the same exposure. Safeonweb's public advice distinguishes general device protection from the particular commercial incidents. [S27]
6.4 Recent activity: 9–22 November
The recent window is an observation period, not a forecast. The CCB's 9 November identity report describes current defensive priorities using earlier reporting populations. Its publication date does not move those underlying incidents into November. [S5]
On 12 November, the tax administration's notice recorded another application restored after the October incident. On 13 November, Reuters recorded attribution uncertainty around refrigeration. Neither is a new independent attack count. [S23][S26]
Safeonweb's 20 November gift-voucher warning adds a customer-data concern. It reports possible disclosure and an ongoing investigation, with no Belgian headcount. Treat this as a trigger for scoped customer communication. [S29]
The recent underground evidence is assessed separately in Chapter 4. A new advertisement can concern old data. It can also promote a tool whose actual use has not been established. Post dates measure the advertisements, not their underlying incidents. [S3]
6.5 Decisions during a short disruption
Agree the minimum information needed to activate traffic protection, isolate a supplier integration or switch to a manual process. Waiting for attribution can prolong disruption without improving the initial containment decision.
Use one public status channel and update it with explicit timestamps. Say which services work, which do not and what users should do. Describe data access only when evidence supports it.
Preserve the distinction between a precaution and a finding. An organisation may isolate a system because it cannot yet rule out risk. That is not a statement that every connected system was affected.
Key takeaways
- 01
Measure service availability separately from criminal claims.
- 02
Operations teams must own safe equipment recovery.
- 03
Recent publication dates do not turn older incidents into new attacks.
Chapter 7
Threat actors and evidence relationships
This chapter answers
Which actor labels are observed, and what relationship can each item actually support?
In short
The evidence contains extortion labels, advertising personas and campaigns described by authorities. These are different kinds of identity. A common sector, language or political theme does not link operators. The report's relationship map connects evidence to a proposition, rather than joining unrelated groups. Build the response around validated access, service impact and recovery needs; use attribution when it improves those decisions.
7.1 How the evidence connects
- Own post
- Relay
- Official statement
- Public reporting
- Our observation
No evidenced link to other actors
The map shows a retained advertisement supporting its own claim, a listing supporting an extortion assertion and public sources supporting a stated outcome. It does not infer a shared operator between those populations.
Matching titles can establish similar advertising. A repeated alias can establish use of that persona. Neither establishes identical files, a completed transfer or the identity of a person behind the account. No common criminal organisation is inferred from these similarities. [S3]
The relationship between reused credentials, account abuse and fraud is an analytical model of a possible pathway. The police describe that pathway in specific investigations. It does not mean the forum sellers in this report supplied those investigated campaigns. [S31][S32]
7.2 Extortion: a distributed listing population
- 1 January-22 November 2026
- Full year 2025
Unmatched periods; not a national risk ranking. Source: Brandefense Ransompedia accepted country-labelled listings, 1 January-22 November 2026; November partial. Listing dates are not incident dates [S1].
Comparison countries were selected as examples to provide context only; they do not imply a ranking or political assessment.
| Country | 2026 window | Full 2025 | Comparison limit |
|---|---|---|---|
| Belgium | 46 | 42 | Listings, not confirmed intrusions |
| Netherlands | 77 | 44 | Neighbouring logistics and Dutch-language market |
| France | 295 | 182 | Neighbouring French-language and enterprise market |
| Germany | 457 | 311 | Neighbouring industrial and trading economy |
| Luxembourg | 2 | 8 | Neighbouring financial-services market with a small listing population |
All 2026 figures cover the same report window. They are not adjusted for business population, economic size, reporting propensity or source visibility. The table cannot establish that a Belgian organisation is safer or more exposed than a peer. [S1]
- Belgian-labelled listings
Source: Brandefense Ransompedia accepted country-labelled listings, 1 January-22 November 2026; November partial. Listing dates are not incident dates [S1].
The monthly listing counts are 1, 2, 5, 7, 3, 4, 3, 7, 4, 5 and 5 from January through the November cutoff. Their sum is 46. Partial November must not be compared with a full month as a trend forecast. [S1]
- Belgian-labelled listings
Source: Brandefense Ransompedia accepted country-labelled listings, 1 January-22 November 2026; November partial. Listing dates are not incident dates [S1].
Qilin has 9 listings, The Gentlemen 7, SafePay 4 and Coinbase Cartel 4. Together they account for 24 of 46 listings. The remaining 22 listings are distributed across 18 other labels. [S1]
Qilin
In Belgium, 2026
- 2026
9 · Belgian-labelled listings in the report window; not confirmed intrusions
Claimed targets
Sector and country labels require validation; no victim list published
Seen in public reporting
- [S1] Retained listing aggregate, not independent actor attribution
Our assessment
Prepare recovery and validate any claim without relying on the actor label
ConfidenceHigh confidence in the retained row count; intrusion outcomes unverified
The Gentlemen
In Belgium, 2026
- 2026
7 · Belgian-labelled listings in the report window; not confirmed intrusions
Claimed targets
Sector and country labels require validation; no victim list published
Seen in public reporting
- [S1] Retained listing aggregate, not independent actor attribution
Our assessment
Prepare recovery and validate any claim without relying on the actor label
ConfidenceHigh confidence in the retained row count; intrusion outcomes unverified
The country label is retained as a listing attribute, not accepted as proof of domestic legal jurisdiction. The listing population includes repeated brand descriptions and a foreign operating unit. Possible organisation overlaps remain distinct listing rows; no unique-victim total is asserted. [S1]
7.3 Advertising personas
| Persona | Cards | Observed proposition | Limit |
|---|---|---|---|
| kazutlg | B01 | Telecom-themed customer-data advertisement | Advertiser itself dates the incident to 2025 |
| xmlrpc | B02 | Citizen-data claim with admitted source uncertainty | Government provenance unverified |
| yneov18 | B03 | Training-network data and access claims | Current access and unique-person count unverified |
| DataJava | B04 | Claimed identity-service administration | Proposed table descriptions do not validate access |
| 0xcasper | B05 | Bank-template phishing-panel marketing | Working capability and deployment unverified |
| updap | B06 | Supplier support-ticket data claim | Customer impact and complete possession unverified |
These are retained account aliases. They are not identified individuals, and their names do not authenticate an advertisement. An alias associated with one accurate item would not validate every later claim. [S3]
7.4 Espionage and unresolved intrusion reporting
The CCB discusses state-sponsored espionage as part of the threat environment. A reported incident at a sensitive public agency is not independently confirmed in the retained official record. The report therefore uses a broad role and attributes the account to reporting. [S8][S9][S14]
The CCB's third-quarter report describes isolated activity that it links to Russia, China and Iran. It also reports a Teams spear-phishing campaign observed in Belgium. These are the authority's assessments. [S9]
The public vendor and NVD records for two mobile-management vulnerabilities establish a product risk. They do not establish which vulnerability, actor or data outcome applied to that reported Belgian incident. [S41][S42][S43]
| Record | Affected scope | Vendor's resolved versions | Boundary |
|---|---|---|---|
| CVE-2026-1281 | NVD lists versions through 12.5.0.0 and specific later vulnerable builds including 12.5.1.0, 12.6.0.0, 12.6.1.0 and 12.7.0.0 | See the release register below | Check the product branch and current vendor guidance |
| CVE-2026-1340 | NVD lists affected EPMM versions through 12.7.0.0 | See the release register below | No inference about an individual Belgian deployment |
Ivanti EPMM fixed releases, as listed by the vendor [S41]:
12.6.1.1, 12.7.0.1, 12.8.0.1Ivanti's advisory describes both as unauthenticated code-injection risks. Its product exclusions include Sentry and its cloud mobile-management product. A remediation decision must match the installed product and branch. The two NVD records were checked for this report. [S41][S42][S43]
No incident-specific ATT&CK matrix is supplied for the unconfirmed agency account. A generic technique mapping would risk turning product capability into an observed Belgian attack sequence.
Key takeaways
- 01
Group labels and personas are not verified operator identities.
- 02
Regional listing totals do not measure comparative attack probability.
- 03
Product vulnerabilities do not independently attribute a reported Belgian incident.
Chapter 8
The fraud economy
This chapter answers
How could advertised data, documents and access support fraud, and where can defenders interrupt it?
In short
Retained advertisements offer identifiers, identity images, sessions and bank-themed tools. Police and Safeonweb describe practical fraud pathways involving impersonation, altered invoices and malicious applications. Those public campaigns are not attributed to the forum personas. Payment-card observations remain a very narrow view. Protect recovery, payment changes and customer communication through independent verification and a clear route for urgent fraud reports.
8.1 From correct details to a false instruction
Real personal information can make a fraudulent message credible. It cannot authenticate the sender. The training, membership and supplier cases explain why customer support and finance need protective controls even when passwords were reportedly unaffected. [S13][S18][S22]
The police's May warning describes criminals using previously leaked mailbox credentials, monitoring incoming messages and altering invoices. It attributes increasing automation to the observed fraud pattern. This report does not independently measure AI's share of that activity. [S31]
The decisive control is independent verification of the instruction. Use an already known contact route before accepting a changed beneficiary or emergency transfer. A familiar mailbox, accurate invoice amount or correct customer name is insufficient by itself.
8.2 Police campaign case cards
Case P
An investigation targets a phishing-panel supplier
- Official account
- Federal Police announced an investigation on 22 April into a suspect alleged to create and sell phishing panels. The statement describes use of those tools in fraud across Belgium and other European countries. [S30]
- Chronology
- The police described a January search and an April transfer of the suspect to Belgium. Those procedural dates differ from the announcement date.
- Established
- The statement supports a crime-service pathway: a supplier offers tools that other criminals can use. It does not connect that suspect to B05 or any persona in the retained forums.
- Limits
- An arrest and police allegations are not a conviction. Do not infer a national victim or loss total from this individual investigation.
- Response
- Make phishing-resistant authentication and payment verification part of the same fraud exercise. Record how a user can report an intercepted authentication interaction.
Case Q
Fake bank staff combine messaging, calls and remote access
- Official account
- A 2 July Federal Police statement describes an investigation involving dozens of Belgian victims and alleged proceeds exceeding EUR 500,000. The investigation was continuing. [S32]
- Chronology
- Police say the investigation began in March and a principal suspect was arrested on 29 June. The July publication is not the start of every reported offence.
- Pathway
- The official account describes messages impersonating public bodies, a subsequent call from a fake bank employee and inducement to install remote-control software.
- Limits
- The proceeds figure is alleged, rather than a national fraud-loss estimate. No common operator with this report's underground advertisements is established.
- Response
- Customer teams should tell users to end unexpected calls and contact the bank independently. Fraud teams need an urgent escalation route for remote-access and new-beneficiary concerns.
8.3 Identity images, sessions and onboarding
The B08 request seeks more than 500 Belgian identity images. It establishes demand only. It does not establish a supplier response, a successful identity check or theft from a government registry. [S3]
The B09 request seeks active sessions associated with logistics services. Sessions can present a different response problem from a password alone. Review active sessions and recovery routes when the organisation has validated a relevant finding. [S3]
Do not turn document possession into sole identity proof. Onboarding and recovery teams should test the circumstances in which a real document image, contact address or old account identifier could be misused. Record an escalation route for uncertain cases.
8.4 Payment-card material in posts: a narrow view
Scope of the card figures
These limited observations do not estimate total card theft, active cards, bank-system intrusion or fraud loss.
The retained material associates 3 distinct payment-card numbers with Belgium, across 9 posts and 17 appearances. No named Belgian issuer attribution is established for these three cards. Repeated appearances are not additional cards. [S3]
These observations cannot establish whether a card remained active or whether any attempted payment succeeded. Issuers should use their own fraud evidence before recommending blocking or reissue. The card population must not be added to credential or customer-record counts.
8.5 Prices are asking prices
| Product advertised | Displayed price | Evidence limit |
|---|---|---|
| Bank-account identifiers | USD 1.75 per 1,000; USD 70.19 for the advertised package | Quantity, provenance and payment authority unverified |
| Claimed identity-service administration | USD 4,550 | Working access and advertised record population unverified |
| Logistics-related sessions | USD 250–2,500 | A buying range, not a completed transaction |
The prices describe the retained advertisements. They do not establish clearing prices, revenue, profitable activity or the size of a Belgian underground market. Differences in claimed scope prevent a simple price comparison. [S3]
8.6 Public campaign warnings throughout the year
| Period | Official warning | Decision supported |
|---|---|---|
| January | High-profile invitation and sponsorship impersonation [S65] | Verify requests through an established business contact |
| February | Government sign-in impersonation [S62] | Open the official service directly |
| March | Fake compensation and recovery messages [S63] | Do not send identity documents to an unverified recovery contact |
| June | Hotel messages with accurate reservation details [S61] | Verify payment requests with the accommodation independently |
| September | Executive impersonation over collaboration software [S60] | Treat a familiar organisational identity as a claim to verify |
| September | Malicious-app and mobile card-relay fraud [S59] | Refuse unsolicited app installation and contact the bank separately |
| October | Messaging-account linking through fake voting requests [S64] | Verify unexpected requests even when they arrive from a known contact |
| November | Fake winter-fuel shops and advance-payment offers [S45] | Check seller identity and payment terms before ordering |
These are prevention and campaign notices. They are not eight disjoint victim populations. The hotel warning, for example, does not establish that the earlier booking-platform incident supplied the information used in those messages. [S61]
Key takeaways
- 01
Correct personal information does not authenticate a payment request.
- 02
A tool advertisement is separate from an observed fraud campaign.
- 03
Card appearances and asking prices do not measure losses or completed trade.
Chapter 9
Banking and finance deep dive
This chapter answers
Which identity and customer-protection signals should the banking portfolio validate first?
In short
11 of 25 selected banking and finance domains had a current credential listing. The sector records at least 13,221 device associations, including 156 employee-classified associations, and 190,956 credential records. The three post-based cards have no named issuer attribution. These observations measure different populations. Coordinate identity, fraud and customer support around validated accounts, sessions and payment activity.
9.1 What the evidence does and does not say
| Population | Available result | Interpretation |
|---|---|---|
| Selected banking and finance domains | 25 | Includes finance services; not a census of licensed card issuers |
| Device associations | At least 13,221; 156 employee-classified | Classification does not establish ownership or an enterprise intrusion |
| Credential records | 190,956 | Overlapping historical sources, not unique active accounts |
| Current credential listings | 11 of 25 domains | Validate freshness, account validity and previous remediation |
| Issuer-domain subset | 7 matched domains; 6 with current listings | 86 employee and 4,306 non-employee device associations; a subset of the sector |
| Bank-template advertising | Observed in B05 | An impersonation product claim, not a bank-system intrusion |
| Payment-card material | 3 cards; 9 posts; 17 appearances | No named issuer attribution; no card-loss or bank-intrusion estimate |
No retained card was attributed to a named issuer. This does not establish the absence of stolen cards. Non-employee associations do not independently establish customer ownership. [S2][S3]
9.2 Separate the account, session and payment decision
An exposed identifier can help personalise a call. A valid session can support account activity. A payment instruction can move funds only within the relevant authorisation process. Investigate those propositions separately rather than treating one as evidence of all three.
When fraud staff receive a report of remote-control software or an unexpected authentication request, route it quickly to account protection. Verify new beneficiaries and recovery changes through an independent known channel. Retain the decision and supporting activity records. [S32]
The B05 advertisement claims bank-specific templates and interactive authentication features. It does not prove that those features work or that any bank's controls were bypassed. Its practical use is to test the fraud team's response to an impersonation scenario. [S3]
9.3 Suppliers and regulatory coordination
DORA has applied since 17 January 2025 to a broad population of financial entities. The NBB describes supervision of ICT and cyber risk, including incident information, tests and third-party dependencies. [S38]
For each essential banking service, identify the identity provider, telecommunications route, core supplier and recovery dependency. A supplier's generic restoration message does not establish that your own service is safe, complete and reconciled.
Maintain one incident chronology and separate decisions for operational reporting, privacy, customer communication and fraud response. The presence of one reporting regime does not automatically close the assessment under another. [S36][S37][S38]
Key takeaways
- 01
Dispose of current listings through account validation.
- 02
Investigate identifiers, sessions and payment authorisation separately.
- 03
Keep issuer subsets, sector associations and post-based cards distinct.
Chapter 10
Regulatory and national response
This chapter answers
Which reporting duties and supervisory milestones should Belgian leaders already have operational?
In short
Belgian NIS2 obligations are already in force. The April 2026 evidence checkpoint and the August supervisory communication require practical preparation, rather than a new-law waiting period. Personal-data incidents also require a separate assessment under GDPR. Financial entities must map the applicable DORA regime. Product manufacturers face CRA reporting duties from September. Assign an owner to each regime and rehearse the first notification before an incident.
10.1 Rules that apply now
The following table distinguishes the trigger, timing and supervisory context. It is a planning aid, not legal advice. The applicable authority and scope depend on the entity, activity and incident. [S33][S35][S36][S37][S38][S39]
| Regime | Applies to | Timing | Responsible route |
|---|---|---|---|
| Belgian NIS2 law | In-scope essential and important entities; significant incidents affecting covered services | Without undue delay after becoming aware of a significant incident: early warning no later than 24 hours; incident notification no later than 72 hours. Final report within one month after notification unless ongoing (see below) | National CSIRT at CCB; trust-service providers have a 24-hour incident-notification exception |
| GDPR | Controllers with a personal-data breach, unless unlikely to create a risk to individuals' rights and freedoms | Without undue delay and, where feasible, within 72 hours of awareness | Competent data-protection authority; APD/GBA is not competent for every sector |
| DORA | Financial entities within the regulation's scope | Applicable since 17 January 2025; use the relevant incident classification and reporting process | Applicable financial supervisor and reporting channel |
| CRA reporting | Manufacturers of products with digital elements and reportable vulnerabilities or severe incidents | Duties began 11 September 2026. After becoming aware of an actively exploited vulnerability or severe incident: early warning within 24 hours and notification within 72 hours | European Single Reporting Platform, relevant CSIRT and ENISA |
If the incident remains ongoing at the one-month milestone, provide a progress report then. Submit the final report within one month after definitive handling of the incident. [S36]
The CRA final-report clocks differ. For an actively exploited vulnerability, the CCB describes a deadline of 14 days after a corrective or mitigating measure becomes available. For a severe incident, it describes one month after the notification. [S39]
Do not apply a deadline from one regime to every incident. Notification to a supplier, insurer or police is not automatically a notification to the competent regulator. Keep an awareness timestamp and a separate decision record for each applicable duty.
10.2 NIS2 conformity: evidence and remediation
The law of 26 April 2024 entered into force on 18 October 2024. The CCB's April communication described evidence expected from essential entities by 18 April 2026. The content depended on the conformity route chosen. [S33][S36]
For CyberFundamentals, the CCB described verification or an active path towards it, including a signed assessment agreement. For ISO/IEC 27001, it requested scope, applicability and internal-audit documentation. Direct inspection required self-assessment and supporting evidence. [S33]
The August communication addresses the 18 April 2027 milestone. Essential entities unable to demonstrate measures equivalent to Essential are asked for evidence equivalent to Important and a remediation plan towards Essential by 18 April 2028. [S34]
No remediation plan is required if, by 18 April 2027, the entity demonstrates Essential-equivalent measures or justifies a lower equivalent assurance level through its risk analysis. [S34]
The CCB explicitly says that this approach does not change the law's obligations or deadlines. Treat the communication as a supervisory requirement to document a credible route, not a blanket extension. Record the route, assessment scope, evidence owner and unresolved gaps. [S34]
10.3 Penalties and exclusions need the right scope
Article 59 distinguishes several breaches of the Belgian NIS2 law. For risk-management and incident-notification duties, it provides different bands for essential and important entities. Other failures have separate bands. [S36]
| Provision | Statutory band | Scope reminder |
|---|---|---|
| Important entities | EUR 500 to EUR 7,000,000 or 1.4% of the undertaking's previous worldwide annual turnover, whichever is higher | Relevant risk-management or incident-notification duties |
| Essential entities | EUR 500 to EUR 10,000,000 or 2% of the undertaking's previous worldwide annual turnover, whichever is higher | Relevant risk-management or incident-notification duties |
| Repeat conduct | The administrative fine doubles for the same facts within three years | Requires the statutory repeat-conduct conditions |
| Public-administration sector | Article 62 excludes these entities from the administrative fines and measures under Articles 59–60 | Assess applicable cybersecurity and notification duties separately [S36] |
These are legal bands, not a prediction of a fine. The law also contains sector-specific exclusions and interaction with equivalent EU rules. Its financial-sector provisions must be read alongside DORA; a generic NIS2 checklist is insufficient. [S36][S38]
10.4 Personal-data notification is a separate decision
The APD/GBA directs controllers to its portal. Its guidance requires notification in a national language; technical annexes may also be in English. It identifies separate competent federal authorities for certain areas, including police information. [S37]
Prepare the portal account and authority assessment before an incident. Record what is known, what remains uncertain and when the next update will arrive. A missing final victim count is a reason to explain uncertainty, not to postpone every reporting decision.
The incident team should give privacy staff a concise impact statement. Include data categories, affected processing, possible recipients, protective measures and the evidence supporting each conclusion. Keep criminal record-count claims separate from the controller's own assessment.
10.5 National crisis coordination
The CCB describes a revised national cyber crisis framework. Its public guidance says the updated plan has applied since 7 May 2026 and coordinates competent authorities without replacing their mandates. Detailed operational content is not public. [S40]
An organisation should still know its own decision chain. Assign authority to isolate a service, request external support, activate manual operations and publish a status notice. Test those decisions with an unavailable supplier and an uncertain data-loss scenario.
Key takeaways
- 01
Keep NIS2, privacy, financial and product-reporting decisions distinct.
- 02
The CCB's remediation approach does not erase statutory deadlines.
- 03
Prepare notification access, evidence owners and crisis decisions before the first alert.
Chapter 11
Outlook
This chapter answers
Which planning assumptions are supported, and which signals would change them?
In short
The strongest planning case is for identity protection, supplier assurance and recoverable services. Official reporting and the cases support those priorities, without establishing a probability of a future incident. Extortion listings remain distributed across many group labels. Recent operational disruption adds urgency to controlling remote access. Use the scenarios below to test decisions, then revise them when your own evidence changes.
11.1 Key judgements
| Judgement | Confidence | Basis | Limit |
|---|---|---|---|
| Account protection deserves immediate attention | High for the priority | CCB identity assessment and police fraud reporting [S5][S31][S32] | Does not measure exposure at an individual employer |
| Supplier failures can become local service problems | High for the pathway | Learning, health and signing-service cases [S8][S13][S22] | Shared dependency does not mean every customer was affected |
| Extortion readiness should cover multiple actor labels | High for the visible listing mix | 46 listings across 22 labels [S1] | Listings do not establish distinct intrusions or losses |
| Remote operational access warrants review | High for the control priority | Refrigeration disruption and official advice [S25][S27] | Does not establish the cause of every reported malfunction |
Confidence belongs to these propositions. It does not certify an actor's honesty, a seller's whole catalogue or the authenticity of every record in a file.
11.2 What would change the assessment
A verified current privileged account would raise the urgency of an organisation's identity response. An old inactive credential with documented remediation would narrow it. The relevant evidence is account validity and recent activity, rather than the volume of advertisements.
A supplier's verified customer-specific impact report would narrow or widen the affected data population. A generic restoration statement would establish availability, while leaving confidentiality unresolved. Document both states independently.
An official investigation or reproducible technical evidence could change attribution for the refrigeration campaign. Repeated allegations, screenshots and political timing would not be enough on their own. Update the judgement only when the proposition itself gains support.
11.3 Conditional scenarios
These scenarios are planning assumptions. They are not forecasts of attack frequency, financial loss or likely perpetrators.
| Scenario | Trigger | First decision | Completion evidence |
|---|---|---|---|
| A current identity is abused | An active account and suspicious session are verified | Identity lead revokes sessions and preserves access records | Account owner confirms recovery; access and recovery paths reviewed |
| A shared supplier isolates its platform | A service notice matches a documented dependency | Service owner activates the tested alternative and requests a scoped impact statement | Manual service works; affected records and reporting decisions documented |
| Remote equipment behaves unexpectedly | Operations validate a setting change or loss of control | Operations lead uses the safe local process and restricts remote access | Safe state confirmed; remote accounts and change history reviewed |
Key takeaways
- 01
Treat scenarios as tests of decisions, not predictions.
- 02
Revise identity urgency using account validity and current activity.
- 03
Change attribution only when stronger evidence supports that proposition.
Chapter 12
Action plan
This chapter answers
Who should act first, and what evidence shows the work is finished?
In short
Begin with facts your organisation can validate: active accounts, exposed services, supplier dependencies and restoration capability. Give each action an owner and a closure record. Extend the exercise to privacy, customer support, finance and operations. The aim is a service that can recover and a team that can explain what happened. Review progress against the evidence below at each leadership meeting.
12.1 The first 30/60/90 days
1Immediate
Remove direct public internet access to refrigeration management interfaces; coordinate safe continuity and protected remote maintenance
- Owner
- Operations and security leads with maintenance providers
- Completion evidence
- Reachability checks confirm no direct public access; safe settings and protected maintenance verified [S25]
230 days
If a current credential or session is linked to your service, verify the account, revoke suspicious sessions and review recovery routes
- Owner
- Identity and SOC leads
- Completion evidence
- Account disposition, activity review and recovery record
Identify internet-facing remote management and business-critical applications; prioritise vendor-confirmed urgent fixes
- Owner
- Infrastructure lead
- Completion evidence
- Owned inventory, patch evidence and documented exceptions
Map critical suppliers to data, administrative access and service processes
- Owner
- Procurement and service owners
- Completion evidence
- Dependency register with an incident contact and alternative process
Prepare one incident decision record covering notification triggers and awareness times
- Owner
- Legal, privacy and incident lead
- Completion evidence
- Tested access to required reporting channels and named deputies
360 days
Rehearse a supplier outage with uncertain customer-data impact
- Owner
- Business continuity and privacy leads
- Completion evidence
- A working alternative, draft user notice and separate regulatory decisions
Test payment-change verification using an already known contact channel
- Owner
- Finance and fraud leads
- Completion evidence
- Sampled callback records and handling for mismatched beneficiary details
Review remote equipment accounts and access paths with maintenance providers
- Owner
- Operations lead
- Completion evidence
- Approved access list, safe local process and change-review evidence
490 days
Restore one essential service from protected backups under realistic dependency constraints
- Owner
- Recovery lead
- Completion evidence
- Measured recovery result, data validation and closed exercise issues
Test detection of suspicious sessions, new authentication methods and unauthorised mailbox forwarding
- Owner
- SOC and identity leads
- Completion evidence
- Reviewed alerts, named response owner and retained exercise evidence
Present the conformity route, supplier gaps and exercise outcomes to management
- Owner
- CISO
- Completion evidence
- Signed decisions, funded remediation and dates for unresolved items
The horizons are organisational planning targets. They do not replace statutory reporting clocks, urgent containment or the immediate response to an active incident.
12.2 Responsibilities by role
Security leadership. Ask for evidence of recovery and identity control, rather than a list of purchased tools. Resolve ownership when an application, supplier and operational service cross several teams. Record accepted residual risks with an expiry date.
SOC and incident response. Start with the account, affected service and evidence type. Retain the original notification and a clear chronology. Verify the difference between a suspicious advertisement, a valid identity and an observed session before describing an outcome.
IT and infrastructure. Check that urgent updates reached the actual exposed system. A successful deployment task does not prove every appliance, bundled runtime or secondary server was covered. Keep recovery access separate from ordinary administration.
Privacy and communications. Draft short notices that say what is known, what remains uncertain and what users should do. Prepare French and Dutch versions together where relevant. Use a stable official channel for later updates.
Finance and customer protection. Verify new payment instructions independently. Train support teams for callers who quote genuine personal information. Real customer details make a fraud attempt more convincing; they do not authenticate the caller.
Operations and suppliers. Define a safe local mode before remote access is restricted. Agree who can change settings, approve emergency access and confirm restoration. Review the maintenance provider's access after staff or contract changes.
12.3 How Brandefense can help
Brandefense provides external threat intelligence and digital-risk monitoring. The service can support investigation of exposed credentials, leaked data and brand impersonation. An alert should enter the validation and response workflow described in this report. [S44]
Bring the domains, brands and services you are authorised to assess to a Brandefense demonstration. Ask how a finding reaches an accountable owner, how uncertainty is represented and what evidence closes the case. [S44]
Key takeaways
- 01
Give every action an owner and a verifiable completion record.
- 02
Test recovery with the supplier and identity dependencies unavailable.
- 03
Keep organisational planning targets separate from immediate incident and notification duties.
Chapter A
Appendix A. Method and limits
This chapter answers
How should a reader interpret the evidence without confusing its populations?
Comparison countries were selected as examples to provide context only; they do not imply a ranking or political assessment.
A.1 Window and timestamp meaning
The report covers 1 January–22 November 2026. Underground observations use post dates, not a presumed intrusion date. Own-source exposure observations date from 21 November; forum and passive observations date from 22 November. The current-day cutoff reflects information available at capture, rather than later hours of that date.
Recent activity is labelled 9–22 November. An official report published during that period can describe earlier events. A restoration update, republication and newly advertised historical file are not new intrusions by default.
A.2 Evidence populations
| Population | Unit | Coverage boundary |
|---|---|---|
| Retained underground material | Distinct post | 1,025 retained forum posts; tagged platform counts reported separately |
| Country-tagged forum subset | Post | 480, included within the retained population |
| Ransomware material | Accepted listing row | 46 Belgian-labelled rows; country and entity scope require caution |
| Domain inventory | Candidate primary domain | 250 selected; 245 with a DNS response |
| Public incidents and campaigns | Evidence dossier | Organisational statements, official warnings and attributed reporting; not a national census |
| Payment-card material | Distinct card and appearances | 3 cards, 17 appearances, 9 posts; no named issuer attribution |
The public chart files contain aggregates. Public evidence cards retain short redacted excerpts without personal samples, victim identifiers or criminal contact routes. An actor alias identifies a forum persona, not a verified person.
A.3 Selected-domain observations
The domain set covers public services and broad economic roles. It is purposive and does not establish a ranked list of Belgium's largest independent organisations. DNS responses do not validate legal ownership, organisational independence or security performance.
Passive email and web observations concern the assessed domains at the observation date. Missing or unsuccessful observations remain distinct from established configuration findings.
Atlas covers all 250 selected domains. The national headline is 48 out of 100. Device, credential, document and current-listing observations were read on 21 November 2026. They are cumulative records, not exclusively 2026 events.
Mentions cover the previous 365 days through 22 November 2026; lookalikes cover 30 days. No organisation or sector rating is published. These observations do not establish active accounts, unique people or confirmed intrusions. [S2]
A.4 Ransomware comparisons
Ransomware leak-site listings are criminal assertions. A listing does not establish encryption, successful theft, payment, loss or unique legal-entity identity. Repeated brand descriptions and foreign units can complicate country comparisons.
The four regional peers share borders with Belgium and offer relevant logistics, language, industry or financial-services context. Raw listing totals are not adjusted national risk rates. Partial 2026 and full 2025 totals are kept explicitly separate.
A.5 Payment-card scope
These limited observations do not estimate total card theft, active cards, bank-system intrusion or fraud loss.
The 3 distinct cards were not attributed to named Belgian issuers. Seventeen appearances are not seventeen cards. The three cards are a single population; there is no additional named-bank population to add to them.
A.6 Public sources and naming
Public source categories include the national cyber authority, Safeonweb, the data-protection authority, police, organisational statements and reporting in relevant languages. Every factual source cited was read for its supporting passage. Verification dates record the review, not the incident date.
Named victims require official confirmation and wider reporting. Other organisations receive broad role labels. Where an article address identifies a withheld organisation, the public source register supplies the publisher's root with the article address withheld.
An unavailable original statement limits naming and corroboration. A reported quotation remains attributed to its speaker and publisher. It is not silently promoted to an independently verified forensic result.
A.7 Confidence and alternatives
High confidence can apply to the existence of a retained advertisement while confidence in its claimed contents remains low. The two propositions are different. Each case records what is held, what is unresolved and an alternative explanation.
Analytical models are labelled. Common language, geography, sector, political rhetoric or publication timing is insufficient to join actors. The report does not infer a shared operator from those similarities.
A.8 What the report excludes
The report does not provide a census of all Belgian cybercrime, a national affected-person total, financial-loss estimates or actor probabilities. Several public notices concern prevention or old events. They are not counted as new attacks merely because they were published in 2026.
For example, an October preventive website shutdown and a November hospital outage without indications of external attack are excluded as confirmed cyberattacks. Their existence illustrates why unavailability alone is insufficient. [S66][S67]
Legal statements are scoped to the cited authority and text. The report is not legal advice. Publication and regulatory dates are distinct from incident dates, and an administrative decision can concern an earlier year's conduct.
Appendix B. Evidence register
| Card | Date | Venue | Alias | Language | Chapter |
|---|---|---|---|---|---|
| B01 | 2026-01-24 | Leak forum | kazutlg | English | 4 |
| B02 | 2026-01-21 | Leak forum | xmlrpc | English | 4 |
| B03 | 2026-10-02 | Leak forum | yneov18 | English | 4 |
| B04 | 2026-11-08 | Leak forum | DataJava | English | 4 |
| B05 | 2026-11-20 | Leak forum | 0xcasper | English | 4 |
| B06 | 2026-11-13 | Leak forum | updap | English | 4 |
| B07 | 2026-04-13 | Leak forum | [alias withheld] | English | 4 |
| B08 | 2026-02-24 | Leak forum | Crazy88 | English | 4 |
| B09 | 2026-11-03 | Leak forum | buylogs | English | 4 |
| B10 | 2026-11-21 | Leak forum | WSdatabase | English | 4 |
Cases A–F assess the retained forum propositions. Cases G–O and R–S assess public incident accounts. Cases P–Q assess official police campaign statements. They are case dossiers rather than mutually exclusive national incident units.
Appendix C. Sources
- S1Odin by Brandefense ransomware-listing aggregate. https://brandefense.io/. Verified on 2026-11-22. Retained report data; URL provides service context, not the underlying records.
- S2Atlas exposure ratings, Brandefense Blackhole domain associations and selected-domain passive observations. https://brandefense.io/. Verified on 2026-11-22. Retained report data; URL provides service context, not the underlying records.
- S3Brandefense Blackhole country-tagged post counts and retained forum and card-material observations. https://brandefense.io/. Verified on 2026-11-22. Retained report data; URL provides service context, not the underlying records.
- S4CCB, Figures show a steady rise in incidents. https://ccb.belgium.be/news/figures-show-steady-rise-incidents-ccb-urges-enterprises-act-immediately. Verified on 2026-11-22.
- S5CCB, The growing threat of identity compromise. https://ccb.belgium.be/news/growing-threat-identity-compromise. Verified on 2026-11-22.
- S6Belgian federal portal, Belgium, a federal state. https://www.belgium.be/en/about_belgium/government/federale_staat. Verified on 2026-11-22.
- S7CCB, Quarterly Cyber Threat Report Q1 2026. https://ccb.belgium.be/nl/open-media/1237/download?inline. Verified on 2026-11-22.
- S8CCB, Quarterly Cyber Threat Report Q2 2026. https://ccb.belgium.be/open-media/1427/download?inline. Verified on 2026-11-22.
- S9CCB, Quarterly Cyber Threat Report Q3 2026. https://ccb.belgium.be/open-media/1470/download?inline. Verified on 2026-11-22.
- S10Security Affairs, hospital service disruption and quoted statement. https://securityaffairs.com/ (article URL withheld). Verified on 2026-11-22.
- S11Safeonweb, company-confirmed fitness and booking-service incidents. https://safeonweb.be/nl/actueel/opgelet-klantgegevens-gelekt-bij-basic-fit-en-bookingcom. Verified on 2026-11-22.
- S12Reuters, Basic-Fit customer-data incident. https://www.reuters.com/business/basic-fit-says-data-breach-exposes-details-200000-members-netherlands-2026-04-13/. Verified on 2026-11-22.
- S13Safeonweb, learning-platform incident and phishing warning. https://safeonweb.be/ (article URL withheld). Verified on 2026-11-22.
- S14RTBF, reported mobile-management incident at a public agency. https://www.rtbf.be/ (article URL withheld). Verified on 2026-11-22.
- S15Machelen, municipal incident updates. https://www.machelen.be/cyberincident. Verified on 2026-11-22.
- S16VRT, municipal confirmation of personal-data theft. https://www.vrt.be/vrtnws/nl/2026/09/23/dan-toch-persoonsgegevens-gestolen-bij-cyberaanval-op-gemeentebe/. Verified on 2026-11-22.
- S17RTBF/Belga, sports membership incident accounts. https://www.rtbf.be/ (article URL withheld). Verified on 2026-11-22.
- S18RTBF, fitness-chain company statement and criminal claim. https://www.rtbf.be/ (article URL withheld). Verified on 2026-11-22.
- S19VRT, school-group incident and director statement. https://www.vrt.be/ (article URL withheld). Verified on 2026-11-22.
- S20VRT, vocational-training data republication. https://www.vrt.be/ (article URL withheld). Verified on 2026-11-22.
- S21L’Avenir, reported public e-learning platform incident. https://www.lavenir.net/ (article URL withheld). Verified on 2026-11-22.
- S22VRT, digital-signature supplier statement and customer-scope distinction. https://www.vrt.be/ (article URL withheld). Verified on 2026-11-22.
- S23SPF Finances/FOD Financiën, recovery updates. https://finances.belgium.be/fr/Actualites/eservices-disponibles-apres-cyberattaque. Verified on 2026-11-22.
- S24VRT, federal tax administration incident. https://www.vrt.be/vrtnws/nl/2026/10/28/cyberaanval-bij-fod-financien/. Verified on 2026-11-22.
- S25CCB, refrigeration-controller security advisory. https://ccb.belgium.be/advisories/warning-six-new-danfoss-vulnerabilities-refrigeration-controllers-risk-immediate-action. Verified on 2026-11-22.
- S26Reuters, differing ministerial assessments of refrigeration attribution. https://www.reuters.com/world/belgian-ministers-odds-over-irans-alleged-role-refrigerator-cyberattacks-2026-11-13/. Verified on 2026-11-22.
- S27Safeonweb, refrigeration incidents and connected-device advice. https://safeonweb.be/nl/actueel/na-de-incidenten-met-koelkasten-warenhuizen-kan-jouw-slimme-koelkast-een-eigen-leven-gaan. Verified on 2026-11-22.
- S28Safeonweb, possible benefits-service data exposure. https://safeonweb.be/ (article URL withheld). Verified on 2026-11-22.
- S29Safeonweb, possible gift-voucher customer-data exposure. https://safeonweb.be/ (article URL withheld). Verified on 2026-11-22.
- S30Federal Police, alleged phishing-panel supplier investigation. https://www.police.be/5998/fr/actualites/crime-as-a-service-arrestation-dun-fournisseur-de-panneaux-de-phishing. Verified on 2026-11-22.
- S31Federal Police, automated invoice-fraud warning. https://www.police.be/5998/fr/actualites/factures-falsifiees-la-police-judiciaire-federale-danvers-tire-la-sonnette-dalarme. Verified on 2026-11-22.
- S32Federal Police, phishing and fake-bank-employee investigation. https://www.police.be/5998/fr/actualites/un-chef-dune-bande-dhameconneurs-arrete-son-retour-de-dubai. Verified on 2026-11-22.
- S33CCB, NIS2 evidence milestone in April 2026. https://ccb.belgium.be/news/nis2-18-april-2026-deadline-what-essential-entities-must-have-place. Verified on 2026-11-22.
- S34CCB, New communication for NIS2 essential entities. https://ccb.belgium.be/news/new-communication-nis2-essential-entities. Verified on 2026-11-22.
- S35CCB, NIS2 Notifications – Howto?. https://ccb.belgium.be/cert/report-incident/nis2-notifications-howto. Verified on 2026-11-22.
- S36Justel, Belgian NIS2 law, consolidated text. https://www.ejustice.just.fgov.be/eli/loi/2024/04/26/2024202344/justel. Verified on 2026-11-22.
- S37APD/GBA, Notifier et gérer une violation de données. https://www.autoriteprotectiondonnees.be/professionnel/actions/violation-de-donnees-personnelles. Verified on 2026-11-22.
- S38National Bank of Belgium, ITA team and DORA. https://www.nbb.be/en/financial-supervision/digital-resilience-and-cyber-risk/ita-team. Verified on 2026-11-22.
- S39CCB, CRA reporting obligations take effect. https://ccb.belgium.be/news/cyber-resilience-act-european-reporting-obligations-take-effect. Verified on 2026-11-22.
- S40CCB, National Cyber Emergency Plan. https://ccb.belgium.be/vital-sectors/national-cyber-emergency-plan. Verified on 2026-11-22.
- S41Ivanti, EPMM security advisory and resolved versions. https://hub.ivanti.com/s/article/Security-Advisory-Ivanti-Endpoint-Manager-Mobile-EPMM-CVE-2026-1281-CVE-2026-1340. Verified on 2026-11-22.
- S42NVD, CVE-2026-1281. https://nvd.nist.gov/vuln/detail/CVE-2026-1281. Verified on 2026-11-22.
- S43NVD, CVE-2026-1340. https://nvd.nist.gov/vuln/detail/CVE-2026-1340. Verified on 2026-11-22.
- S44Brandefense, Request a demo. https://brandefense.io/request-a-demo/. Verified on 2026-11-22.
- S45Safeonweb, fake pellet and firewood offers. https://safeonweb.be/nl/actueel/pellets-brandhout-kopen-laat-je-niet-misleiden-door-valse-aanbiedingen. Verified on 2026-11-22.
- S46Charging-account service, company incident FAQ. https://help.chargemap.com/nl/articles/11583170. Verified on 2026-11-22.
- S47Chemical manufacturer, clarification of the May incident. https://www.christeyns.com/ (article URL withheld). Verified on 2026-11-22.
- S48Recycling operator, forensic follow-up. https://www.iok.be/ (article URL withheld). Verified on 2026-11-22.
- S49Medical research institute, statement updated in November. https://www.itg.be/ (article URL withheld). Verified on 2026-11-22.
- S50Professional legal body, scope update. https://www.avocats.be/ (article URL withheld). Verified on 2026-11-22.
- S51ITdaily, supermarket supplier incident. https://itdaily.be/ (article URL withheld). Verified on 2026-11-22.
- S52Wonen360, furniture-retailer company statement. https://www.wonen360.nl/ (article URL withheld). Verified on 2026-11-22.
- S53Télésambre, municipal incident and restoration reporting. https://www.telesambre.be/ (article URL withheld). Verified on 2026-11-22.
- S54La Libre, precious-metals customer-data incident. https://www.lalibre.be/ (article URL withheld). Verified on 2026-11-22.
- S55DH, municipal website incident and scope. https://www.dhnet.be/ (article URL withheld). Verified on 2026-11-22.
- S56HLN/VTM, travel-operator statement and customer messages. https://www.hln.be/ (article URL withheld). Verified on 2026-11-22.
- S59Safeonweb, malicious-app and NFC-relay warning. https://safeonweb.be/nl/actueel/pas-op-voor-frauduleuze-telefoontjes-waarin-je-wordt-gevraagd-een-app-te-installeren-en-je. Verified on 2026-11-22.
- S60Safeonweb, executive impersonation on Teams. https://safeonweb.be/nl/actueel/pas-op-voor-ceo-fraude-microsoft-teams. Verified on 2026-11-22.
- S61Safeonweb, false hotel and B&B reservation messages. https://safeonweb.be/nl/actueel/opgelet-voor-valse-whatsapp-berichten-over-hotel-en-bb-boekingen. Verified on 2026-11-22.
- S62Safeonweb, government sign-in impersonation. https://safeonweb.be/nl/actueel/phishing-naam-van-csam. Verified on 2026-11-22.
- S63Safeonweb, fraudulent compensation and recovery messages. https://safeonweb.be/nl/actueel/opgelet-voor-fake-dossier-voor-schadevergoeding-na-fraude. Verified on 2026-11-22.
- S64Safeonweb, false voting and account-linking messages. https://safeonweb.be/nl/actueel/opgelet-voor-valse-stemwedstrijden-whatsapp. Verified on 2026-11-22.
- S65Safeonweb, high-profile invitation impersonation. https://safeonweb.be/nl/actueel/opgepast-voor-berichten-die-van-het-koningshuis-lijken-te-komen. Verified on 2026-11-22.
- S66Safeonweb, preventive PHP patching outage. https://safeonweb.be/nl/actueel/php-crisis-waarom-waren-verschillende-websites-niet-bereikbaar. Verified on 2026-11-22.
- S67Numerikare/Belga, hospital outage without external-attack indications. https://www.numerikare.be/ (article URL withheld). Verified on 2026-11-22.
- S68IETF, RFC 6797: HTTP Strict Transport Security. https://www.rfc-editor.org/rfc/rfc6797.html. Verified on 2026-11-22.
- S69W3C, Content Security Policy Level 3, abstract. https://www.w3.org/TR/CSP3/. Verified on 2026-11-22.
- S70IETF, RFC 4034: Resource Records for the DNS Security Extensions, section 5. https://www.rfc-editor.org/rfc/rfc4034.html. Verified on 2026-11-22.
- S71VRT, charging-account service incident and European user population. https://www.vrt.be/vrtnws/nl/2026/11/18/charge-map-cyberaanval/. Verified on 2026-11-22.
Appendix D. Glossary
| Term | Plain meaning |
|---|---|
| Account recovery | The process for restoring access after credentials or authentication methods are lost |
| Credential | Information used to authenticate, such as a password or authentication secret |
| Device association | A record linking a device-related observation to a domain; ownership requires validation |
| DMARC | An email policy that helps receivers assess messages using a domain's identity |
| DKIM | A digital email signature checked against a domain's published verification information |
| DNSSEC | A system for validating signed DNS information |
| DS record | A delegation signer record links a parent DNS zone to a child zone’s signing key; observing it does not validate the full DNSSEC chain [S70] |
| HSTS | HTTP Strict Transport Security directs browsers to use HTTPS for a site; one observed header does not assess every application [S68] |
| CSP | Content Security Policy controls which resources a page may load or execute; one observed policy does not establish complete application security [S69] |
| Exfiltration | Unauthorised copying of data out of a system |
| Initial access broker | A seller who advertises access to organisational systems; each access claim needs validation |
| Metadata | Information describing another record, such as its title, author or processing time |
| NFC relay | Forwarding short-range card communications to another location during a fraudulent transaction |
| OAuth consent | Permission allowing an application to access specified resources under an account |
| Pseudonymisation | Replacing direct identifiers while keeping a possible route to linking records to people |
| Session | An authenticated interaction that may remain usable after the initial login |
| SPF | A DNS-published statement of authorised sending services for an email domain |
| TLS | The protocol used to protect network communication and authenticate a service |
About this report
Prepared by Odin by Brandefense. Brandefense Country Report · Belgium 2026. Version 1.0, observation cutoff 22 November 2026. TLP:CLEAR.
The report keeps advertisements, listings, official accounts and assessment separate. It takes no political position and makes no independent state attribution. Recommendations describe actions an organisation should validate; they do not assert that a named organisation has failed to perform them.
French and Dutch editions are translated with AI from the English master. The English edition prevails where they differ. Independent publication review, rendering and release are separate steps.
© 2026 Brandefense. All rights reserved.