2012 · United States · J - Information and communication · Credential theft and later resale of password data
Analysed 2026-11-06 by Hermes
Open this case in the tool →Credential theft and later resale of password data
Expertise to Execution
Evidence coverage: 3/6 stages, weight 15/60
Decisive stage: egress (clearest evidence) · Alternative: Strategy to Leadership
(1) The expectation at a professional network holding hundreds of millions of member accounts was that the crown-jewel credential store be protected to the standard of the day - salted, strong password hashing - and that a theft of that store be scoped and its affected members remediated. (2) In practice, the 2012 theft exfiltrated a password file still hashed with unsalted SHA-1; LinkedIn presented hashing and salting as security it had only just put in place in June 2012 and confirmed in 2016 that it now uses salted hashes; the full scale of the theft (some 117-164 million accounts against the 6.5 million treated in 2012) came to light only on 17 May 2016 when the data was offered for sale, and the 2016 remediation again followed a data-matching rule that missed some affected members. (3) That is a knowing-doing gap inside the material column, not generic weak security: the remedy was known, standard and demonstrably implementable - LinkedIn deployed salting right around the incident - yet production password storage did not realise it until after the theft (material origin, execution break), and remediation reach was limited by operational data matching. No source shows leadership deferring, underfunding or rejecting the hashing remedy, so no doctrinal-column gap is read; the entry vector of the 2012 theft was never disclosed and is not load-bearing. The classification rests on the egress stage (the crown jewels were not protected differently), the stage whose origin and break are best documented; the detection and response gaps point the same way. (4) RM26 (Expertise to Execution lens): LinkedIn's password architecture remained unsalted SHA-1 because adequate hashing and key protection for the credential store had not been implemented in production, and salting was introduced only around the time of the theft, so the stolen hashes were plaintext-equivalent and cracked at scale (causal role: Amplifying). RM34 (Expertise to Execution lens): available monitoring and breach-scoping practice was not configured or maintained to reveal the theft's extent - the enlarged dataset surfaced only through an external marketplace listing four years later (causal role: Obscuring). RM43 (Expertise to Execution lens): specialists could not fully identify or reach affected parties because remediation relied on matching leaked-data attributes rather than the 2012 membership record, so some affected members were never notified (causal role: Amplifying).
RM26 — Data-protection / cryptographic weaknessRM34 — Monitoring / detection / alert-handling gapRM43 — Affected-party remediation gapThe case file's description matches its stated victim (LinkedIn, credentials stolen in 2012, discovered on the black market in 2016) - no mismatch. Both source_url leads (CNN 2016 and CNN 2012 money.cnn.com articles) could not be opened (HTTP 503 / fetch failures) and were replaced by the sources listed under Manual source-article; no Wayback snapshot of them was reachable from this environment (CDX and availability API returned nothing), so no archive URL is cited. The LinkedIn May 2016 blog post 'Protecting our members' cited by the dataset is no longer at its original URL; its text is quoted by the Krebs on Security and Infosecurity Magazine articles cited here, and LinkedIn's current 2016 help-centre breach notice carries the same account. Leaked Source / marketplace figures (167M records, 90% cracked in 72 hours) are third-party claims as reported by cited press and are used only where corroborated (KoreLogic's hash counts, Hunt's cracking analysis). Actor is unattributed; the theft method was never disclosed. The related row 893 (listed in the case file for the same victim) has no case file in the run's cases folder, so it could not be opened or compared, and nothing is stated about it here. No other incident at LinkedIn was substituted for the described one.