2021 · Philippines · O - Public administration and defence · Cyber-espionage intrusion via self-replicating USB-borne malware
Analysed 2026-11-07 by Hermes
Open this case in the tool →Cyber-espionage intrusion via self-replicating USB-borne malware
Expertise to Execution
Evidence coverage: 3/6 stages, weight 19/60
Decisive stage: detection (clearest evidence) · Alternative: Strategy to Execution
(1) No named victim left a doctrinal statement to measure against: the affected parties are an unnamed set of public and private entities in the Philippines, so the expectation that applies is the operational baseline that specialist practice prescribes for this exact threat - controlling execution from removable media, monitoring endpoints for signed-binary side-loading, Run-key persistence and NCAT reverse shells, and retaining logs long enough to reconstruct activity. Mandiant's own remediation of the campaign (new real-time detections, YARA rules and hunt logic for these behaviours) shows these remedies were available and effective. (2) In practice the baseline did not exist in the victim environments: attacker-planted USB drives were plugged in and their renamed signed binaries manually executed, DARKDEW then re-infected every new drive connected, letting the payloads propagate across additional systems and potentially air-gapped ones, and the activity sat undetected from at least September 2021 until Mandiant's proactive threat hunting surfaced it in late November 2022 - the TTPs could evade traditional detection mechanisms, and Mandiant could not even confirm reverse shell interaction because of visibility gaps or short log retention periods. (3) That is a knowing-doing gap inside the material column: the remedies were standard technical expertise, not policies or leadership decisions (nothing in the record documents a victim policy, deferred programme or discounted warning, so no doctrinal or leadership break can be evidenced), yet they were not realised in the operational environment for roughly 14 months; the classification rests on the detection stage, the only gap whose origin and break are both documented in a primary forensic source; the entry vector is fully documented but pattern-level for an unnamed victim set, and containment (USB worming) reinforces rather than redirects the same material-execution reading. RM25 (Expertise to Execution lens): media controls were not implemented in the victim environments - attacker-supplied USB drives were carried in, plugged in and manually executed, and every newly connected drive was silently re-infected - standard media hardening never realised operationally (causal role: Enabling). RM34 (Expertise to Execution lens): the available monitoring did not detect the intrusion: the TTPs could evade traditional detection mechanisms and the campaign ran from at least September 2021 until discovery by an external party's proactive threat hunting in November 2022, a roughly 14-month dwell none of the affected entities' own alert handling shortened (causal role: Obscuring). RM33 (Expertise to Execution lens): the incident could not be reconstructed: Mandiant could not observe any reverse shell interaction given the age of the activity, which it attributes to visibility gaps or short log retention periods - the log and visibility tooling needed to investigate was not in place (causal role: Obscuring).
RM25 — Physical / media security gapRM33 — Logging / forensic-readiness gapRM34 — Monitoring / detection / alert-handling gapVictim is a diffuse, unnamed set of public and private entities (receiver_category Unknown in the dataset); named only descriptively per the dataset's convention, so no organisation-level lifecycle detail (own policies, leadership decisions, response conduct) exists anywhere in the record - victim-side conditions are read from Mandiant's forensic description and coded with that caveat. Sector coded O Public administration and defence following the convention for multi-entity campaigns with public entities among the primary targets (consistent with other diffuse-victim rows in this run); the campaign also hit private entities, so no single-sector reading is exact. Country from the dataset's receiver_country list (also US, Europe, Oceania and Northeast Asia appear; the Philippines is the main target per Mandiant and the victim description). Year 2021 from eurepoc_extra.start_date (01.09.2021, backed by PE compile timestamps); public disclosure 28-30.11.2022 by the IT-security company. Attribution rests solely on Mandiant's technical assessment (China nexus); no regulator or victim confirmation exists. Both current mandiant.com URLs from the case file now redirect to a marketing homepage, so the report is cited via the Internet Archive snapshot opened and read during this task. The egress-stage uncertainty (no reverse shell interaction observed; short log retention) is coded as RM33 rather than RM35: exfiltration is assessed as the campaign's purpose but not incident-confirmed, so no egress gap is recorded. weighted_intensity 1 / incident_type Hijacking without Misuse per the dataset.