← Breach analysis by the Dual Vulnerabilities framework

Sejong Institute watering hole attack (Andariel, 2018)

2018 · South Korea · M - Professional, scientific and technical activities · Watering hole attack (website compromise with ActiveX zero-day against site visitors)

Analysed 2026-11-07 by Hermes

Open this case in the tool →

What happened

Watering hole attack (website compromise with ActiveX zero-day against site visitors)

Dual vulnerability type

Expertise to Execution

Evidence coverage: 1/6 stages, weight 6/60

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

Why this classification

(1) The strategic expectation at a national-security think tank whose public website is read by the security-community audience it targets is baseline integrity and visibility of that web presence: an organisation should know what code its own site serves and should notice injected files - whether this expectation was institutionalised in any institute policy is nowhere documented. (2) In practice the attackers operated on the victim's own website with near-total impunity for about sixteen months: a reconnaissance script was planted around January 2017, the ActiveX zero-day exploit was injected at least three times during 2018 timed around the April Inter-Korea summit, and nothing on the victim side registered any of it - AhnLab and IssueMakersLab found the compromise in May 2018 and AlienVault disclosed it on 11 June 2018, while the route by which the website was first altered and the institute's own response remain undisclosed. (3) This is a knowing-doing gap within the material column: the remedy that would have revealed each injection - file-integrity and web-traffic monitoring of the public web server - is standard specialist knowledge, yet it was not executed in operations across three separate injections over a year and a half; the classification rests on the detection stage (the only stage with a documented victim-side gap; the entry route is not load-bearing because it is undocumented), origin material (no doctrinal artefact of the victim is documented) and break execution (no source shows any leadership decision, deferral or resourcing failure). A secondary doctrinal reading (Strategy to Execution) is visible but cannot be placed on evidence because no victim-side policy is documented. (4) RM34 (Expertise to Execution lens): the institute's available detection capability was absent or never configured and maintained to notice altered and malicious files on its own public web server, so the compromise persisted roughly 16 months and surfaced only through external researchers rather than any internal alert (causal role: Obscuring). RM15 (Expertise to Execution lens): the technical teams demonstrably lacked an accurate picture of the content and exposure path of their own web assets - attacker scripts hidden inside an ordinary jQuery file path were injected and re-injected three times without any internal visibility (causal role: Enabling). The exploited ActiveX control was a zero-day at the time of attack, so no patch-latency marker is coded; the government-mandated enablement of ActiveX that enlarged the exposed visitor population is a national environmental condition, not a victim-side marker, and is described here rather than coded.

Evidence by stage

Risk markers

Sources

Notes

The case file names the victim only as a South Korean think tank focused on national security; secondary reporting (ZDNet, Dark Reading) and the sejong.org indicator in the AlienVault report identify it as the Sejong Institute. The case file's source URL to cybersecurity.att.com (AlienVault post after the AT&T acquisition) returns HTTP 502 and was replaced with the original AlienVault forensic post read via a Wayback snapshot dated 2018-06-15. EuRepoC's start_date 12.06.2018 coincides with AlienVault's disclosure (the post is dated 11 June 2018), while the compromise itself ran from about January 2017 (reconnaissance) with zero-day injections in April-May 2018; Year 2018 holds either way. No victim statement, regulator action or Korean-language reporting of a Sejong Institute response was found, so the response stage is recorded as undocumented rather than aligned; disclosure was by IT-security companies. Zero-day status excludes RM18. Korean think tank sector coded as M (professional, scientific and technical activities). No related_rows in the case file.