2020 · United States · J - Information and communication · Hack with mass privacy-settings reset across a DNA-profile database (two intrusions)
Analysed 2026-11-07 by Hermes
Open this case in the tool →Hack with mass privacy-settings reset across a DNA-profile database (two intrusions)
Expertise to Execution
Evidence coverage: 4/6 stages, weight 23/60
Decisive stage: response (clearest evidence) · Alternative: Expertise to Leadership
(1) GEDmatch held the genetic profiles of about 1.4 million people, and after Verogen's December 2019 acquisition its business rested on the explicit promise that user data would be protected and the law-enforcement opt-in boundary enforced; outside security researchers had flagged concrete weaknesses from 2019 onward. (2) In practice an intruder inside a single existing user account reset the privacy settings of every account in one action (all opted-out profiles law-enforcement-searchable for about 3 hours, hidden law-enforcement kits visible to all users); the first remediation did not hold and the site was re-breached on 20 July; and within 48 hours attackers were phishing MyHeritage with GEDmatch users' names and email addresses while the company's public position was that no user data was downloaded or compromised. (3) The classification rests on the response stage: the remedy - removing the exploited vulnerability before reopening the site - was known, attempted and resourced, and the chain broke in realising it: Verogen's own statement records that the site was still vulnerable when the second attack came. That is a knowing-doing gap inside the material column (expertise existed and was funded but was not executed), not a leadership deferral - no source shows a decision or budget withheld from the fix - and not a doctrinal-design gap. The entry vector (how the account itself was compromised) was never disclosed and is not load-bearing; containment (unbounded blast radius of one account) and egress (no ability to scope what was taken) point the same way, and the second DV pattern, Expertise to Leadership, sits in the pre-breach condition where academic warnings were discounted. (4) RM38 (Expertise to Execution lens): responders could not fully eradicate the intrusion - the first fix left the server exploitable, the attacker returned the next day and the incident's impact doubled (causal role: Amplifying). RM30 (Expertise to Execution lens): the architecture gave one ordinary user session write reach over the privacy settings of the entire database with no tiering between user kits and law-enforcement kits, so the blast radius was all 1.4 million profiles (causal role: Amplifying). RM35 (Expertise to Execution lens): egress visibility was never built even though researchers had shown mass harvesting looks like ordinary relative searches, so GEDmatch could not confirm or refute what the intruder took and could only assert no data was downloaded (causal role: Amplifying). RM07 (Expertise to Execution lens): academic warnings about specific weaknesses reached the founders but were publicly downplayed rather than made actionable as funded hardening before the attack (causal role: Obscuring).
RM38 — Containment / eradication failure during responseRM30 — Segmentation / containment weaknessRM35 — Egress / exfiltration monitoring gapRM07 — Risk signal / escalation suppressioninformationisbeautiful row; the dataset description ('DNA data on up to 1.4m users... may have been hacked') matches the case but overstates the confirmed theft: reporting confirms ~1.4m profiles were made law-enforcement-searchable by two permission-reset hacks on 19-20 July 2020; Verogen's own disclosure says no user data was downloaded or compromised, and theft of the DNA archive was never confirmed - MyHeritage only suspects the attackers took names and email addresses (used in targeted phishing from 21 July, 105 users lured, 16 entered credentials). Entry vector limited to 'an existing user account'; how that account was compromised is undisclosed. Disclosure timeline: Facebook statement 20 July, user email 22 July; on 22 July GEDmatch still said there was no evidence the MyHeritage phishing scam connected to its breach, one day after MyHeritage published the linkage evidence. Cruwys blog notes a GDPR reporting duty for EU users; no regulator action against GEDmatch was found in the reporting reviewed. No related_rows in the case file. Link check: nytimes.com blocks automated access (HTTP 403); the page was opened and read during research (NYT text also published verbatim by the Boston Globe).