2012 · United States · J - Information and communication · Data theft via SQL injection (plaintext credential dump)
Analysed 2026-11-06 by Hermes
Open this case in the tool →Data theft via SQL injection (plaintext credential dump)
Expertise to Execution
Evidence coverage: 3/6 stages, weight 20/60
Decisive stage: egress (clearest evidence) · Alternative: Strategy to Execution
(1) The expectation at Yahoo sat on both axes: publicly it claimed it takes security very seriously and invests heavily in protective measures, and at the technical level every remedy relevant here was basic, known expertise for a major web company - input validation against SQL injection (a well-known attack), hashed and salted password storage, and intrusion-detection trip-wires, which an expert said Yahoo's own policies for its other sites should have implied. (2) In practice the acquired Associated Content / Yahoo Contributor Network property ran on a SQL-injectable web application and stored roughly 453,000 credentials in plaintext, the data dating from a pre-May-2010 legacy file that was never migrated or protected after the 2010 acquisition, and the theft surfaced only when the attackers published the dump on 11-12 July 2012. (3) That is a material-to-execution gap: the remedies were known at specialist level and needed no new decision or funding, but were simply never realized in this property's operations and architecture; no source shows leadership deferring or declining them (a leadership break is unevidenced), and no doctrinal artefact specific to the property is documented, so the doctrinal reading stays runner-up. The entry vector is documented but not load-bearing for the type - it shows the property was reachable; the classification rests on the egress stage (credential protection), corroborated by the entry gap, with the detection stage (a Strategy to Execution pattern, per the site-wide policy comment) contributing. (4) RM29 (Expertise to Execution lens): secure-development practice against SQL injection is basic developer knowledge, yet the acquired Contributor Network application was never built or hardened with it, leaving a union-based SQL injection that handed over the database (causal role: Enabling). RM26 (Expertise to Execution lens): around 453,000 user credentials sat in plaintext with no hashing or salting, the standard protection experts call routine, so the stolen file was instantly usable and the leak's impact was amplified across Gmail, Hotmail and AOL accounts (causal role: Amplifying). RM20 (Expertise to Execution lens): the 2010 acquisition of Associated Content was integrated without the legacy account base being hardened - the pre-acquisition plaintext file survived inside Yahoo's infrastructure until 2012 and the accounts were not brought into the stronger Yahoo login system, as a Cambridge researcher criticized (causal role: Enabling).
RM29 — Secure-development / non-production exposureRM26 — Data-protection / cryptographic weaknessRM20 — Change / transition control failureCase-file source_url (helpnetsecurity.com 2012-07-12 article) could not be opened: the live page returns HTTP 405 behind a CAPTCHA bot wall and the Wayback CDX index holds no snapshot; the analysis rests on the contemporaneous articles opened instead. Link check: hartfordbusiness.com blocks automated access (HTTP 403); the page (a full republication of the CNN Money article of 12 July 2012 by David Goldman) was opened and read during research. The Computerworld report on Yahoo's follow-up statement now returns 404; the CNET FAQ is cited instead and quotes both Yahoo statements verbatim. The dataset description says 'Yahoo's Voice service': the affected service was Yahoo Voices / the Yahoo Contributor Network (formerly Associated Content); SecurityWeek notes media conflated the separate Yahoo Voice product. Attribution to D33Ds Company rests on the group's own claims as reported by the press; Yahoo did not name an attacker. The affected data predated Yahoo's May 2010 Associated Content acquisition per Yahoo's own statement; no earlier comparable, documented Yahoo incident was verified from opened sources, so no recurring-weakness marker (RM21) was coded despite the attackers' unverified claim that Yahoo webservers had been exploited before.