← Breach analysis by the Dual Vulnerabilities framework

Ticketcounter Data theft & extortion (cybercrime trio, 2021)

2021 · Netherlands · J - Information and communication · Data theft via exposed cloud storage, followed by extortion

Analysed 2026-11-06 by Hermes

Open this case in the tool →

What happened

Data theft via exposed cloud storage, followed by extortion

Dual vulnerability type

Expertise to Execution

Evidence coverage: 4/6 stages, weight 26/60

Decisive stage: entry (clearest evidence) · Alternative: Strategy to Execution

Why this classification

(1) Ticketcounter processes reservations and payments for zoos, theme parks and museums, so under the AVG/GDPR the personal data of its partners' customers (names, dates of birth, addresses, phone numbers, IBANs) had to be protected and minimised; the company was itself standing up an anonymization process for that data. (2) In practice a copy of the production database - 1.5 to 1.9 million ticket buyers, fully unmasked, covering 2017 to August 2020 - was placed on an Azure container whose public access was still on by default and could be downloaded straight by URL; it sat there from 5 August 2020 until an e-mail from KPN Security on 22 February 2021 showed the data advertised on RaidForums. Bakker refused the 7-bitcoin ransom, reported to the police and informed partners and the AP, but the exposed file had been deleted before preservation, so the exact damage could not be determined from the company's own evidence and the scale was established only through a copy Troy Hunt had retained. (3) The gap is knowing-doing within the material column: the remedies - turning off blob public access / securing the copied container, and masking the copy before externalising it, the very process the copy was made to test - were available, routine technical steps that no source shows were ever executed, and the organisation's own monitoring missed the exposure for roughly six and a half months while an outside security firm caught it. No source documents a Ticketcounter policy that leadership deferred or under-resourced, so the break sits in execution, not leadership; the attacker's technique (finding an exposed container) is context only and not load-bearing for the type. The classification rests on the entry stage, where the origin and break are best documented by the victim's own confirmation; detection, egress and response show the same material-to-execution pattern. RM27 (Expertise to Execution lens): the routine configuration control experts are meant to maintain - private access on a cloud container holding a production database copy - was never applied, the container keeping Azure's then-default public access until an attacker found the URLs (causal role: Enabling). RM34 (Expertise to Execution lens): no monitoring or exposure detection had been configured or tuned to flag a crown-jewel dataset sitting publicly reachable for months, so first detection came externally from KPN Security roughly six and a half months into the exposure (causal role: Obscuring). RM16 (Expertise to Execution lens): the unmasked three-year production copy was handled outside controlled systems without inventory or masking controls, so the anonymization/minimisation duty the copy was made to operationalise never took effect and the theft exposed fully identifiable, unminimised records (causal role: Amplifying). RM33 (Expertise to Execution lens): after the exposed file was deleted, specialists could not reconstruct what had been taken or how much - the exact damage could not be determined - because no evidence had been retained or logged; scope came only from an outsider's copy (causal role: Obscuring).

Evidence by stage

Risk markers

Sources

Notes

Multi-victim EuRepoC row; victim as recorded: 'Not available;Ticketcounter;RDC;Not available;Shell;Not available;Not available'. The DV is assessed on Ticketcounter, the only named victim whose victim-side lifecycle is documented (RDC is an ICT service provider for car garages whose breach hit data of millions of car owners, and Shell was hit via a third party, but no lifecycle detail for them was found in accessible sources). The row itself describes the police investigation and the January 2023 arrests announced 23 February 2023 (case-file description matches the politie announcement); it is not an attacker-side takedown - real victim organisations with documented gaps anchor the classification. District Court Amsterdam, 6 November 2023: main suspect sentenced to four years (one suspended) plus nearly EUR 300k damages; one unnamed victim paid almost EUR 750k - Ticketcounter did not pay. The AP reprimanded 46 partner organisations that failed to notify their own customers (controller-side notification duty, not Ticketcounter's, so not coded). Data-category discrepancy: BleepingComputer's sample showed hashed passwords and IP addresses among the exposed fields, while Dutch reporting (VPNGids, security.nl) said login and credit-card data were not leaked; the DV reading does not depend on which is right. Link check: politie.nl blocks automated access (HTTP 403); the page was opened and read during research. Link check: ib-p.nl blocks automated access (HTTP 403); the page was opened and read during research. The DIVD statement on its former volunteer (divd.nl, 24-02-2023) would not open (http_error); the DIVD detail is therefore cited via the Malwarebytes report instead.