← Breach analysis by the Dual Vulnerabilities framework

Epsilon Email Marketing Breach (cybercrime ring, 2011)

2011 · United States · J - Information and communication · Mass theft of customer names and email addresses via a third-party email marketing platform

Analysed 2026-11-06 by Hermes

Open this case in the tool →

What happened

Mass theft of customer names and email addresses via a third-party email marketing platform

Dual vulnerability type

No DV evidenced

Alternative: Strategy to Execution

Why this classification

(1) Strategic expectation: Epsilon operated the email marketing platform and customer databases of some 2,500 clients, including the largest US banks and retailers, and before the breach maintained, per its own account to the regulator and to Congress, an ISO 27001/27002-conforming information-security program running for five years, annual SAS 70 Type II audits, a board-approved security policy reviewed annually and signed by employees, and annual security training. (2) In practice: malware infected a remote-working employee's workstation while browsing the internet and a second-stage keylogger captured the login credentials for the email marketing platform; the attacker used those credentials to log on between 21 February and 30 March 2011, and on 29 March client information was downloaded to an FTP site in another country; the intrusion surfaced on 30 March only when an employee reported unusual download activity to Epsilon's internal security hotline; Epsilon then disabled the credentials immediately, began forensics within hours, notified clients the same day, published statements on 1 and 6 April and handed the case to the FBI and US Secret Service. (3) Why no dual vulnerability: the Australian Privacy Commissioner conducted an own-motion investigation of exactly this question - whether Epsilon's overall safeguards met the duty to take reasonable steps to protect the data (NPP 4.1) - and concluded Epsilon had taken reasonable steps and was not in breach, attributing the incident to a sophisticated, expertly executed attack rather than a failure of controls; under the method's operational rule a formal investigation finding no failings is the positive evidence of alignment that No DV evidenced requires. No primary source documents a specific gap the regulator did not consider: the malware mechanism, the five-week credential-abuse period and the hotline detection are the facts on which the investigation itself relied, and the 2015 US federal indictment was directed at the attackers' ring, of which Epsilon confirmed it was a victim. The entry mechanism is documented and not load-bearing for a DV reading. Conditions the evidence shows without a DV claim are recorded as risk markers only: a single-factor credential path into the platform holding customer-list crown jewels, and monitoring that did not surface weeks of anomalous use of valid credentials before a human alert. A data-minimisation condition is likewise documented but described rather than coded: Epsilon confirmed in congressional testimony that addresses opted out years earlier were retained solely for CAN-SPAM compliance and were among the records stolen. Recovery was not in play (data theft without service disruption), and the response record is documented as prompt.

Risk markers

Sources

Notes

informationisbeautiful row; the one-sentence description and the cited Guardian article match the March/April 2011 Epsilon incident (the source_url in the case file carries a stray trailing ' /'). Mechanism, dwell period and response taken from the OAIC own-motion investigation report (Dell Australia and Epsilon) and corroborated by the Alberta OIPC findings report (p2011-nd-011), which records the 29 March 2011 download of client information to an FTP site in another country and the 30 March alert investigation. No DV evidenced rests on the operational rule for regulator findings of no failings: the Privacy Commissioner examined Epsilon's safeguards and found no failing; caveats are that the Commissioner did not reach a formal view on the Act's extraterritorial application to Epsilon and that the review focused on NPP 4.1 for Australian data - no equivalent US regulator investigation of Epsilon occurred (the 2015 DOJ prosecution targeted the attacker ring; the indictment does not name Epsilon, though Epsilon confirmed in a statement it is among the victims, per Krebs on Security). The 2010 industry-wide spear-phishing campaign against email service providers (Krebs, Nov 2010) preceded the attack and Walgreens said it had asked Epsilon for more security after the late-2010 incidents (reported via Dark Reading), but this is secondary reporting of a general expectation, not a documented specific gap, so it is not treated as an exception to the regulator finding. Scale across the ESP sector (more than a billion addresses) comes from the DOJ allegations as reported; the per-client scale Epsilon itself published was limited to 'a subset' (about 2%) of clients. Sector: Epsilon's primary activity is operating an email marketing platform and hosting customer databases (information services), coded J. Link check: oipc.ab.ca was temporarily unavailable (HTTP 503) at check time; the page was opened and read during research. Unattended run: no consent-requiring actions were needed; only public sources were read, no leak or attacker infrastructure was contacted.