← Breach analysis by the Dual Vulnerabilities framework

Guyanese government entity data theft (Unspecified China-aligned hacking group, 2023)

2023 · Guyana · O - Public administration and defence · Cyberespionage data theft via spearphishing and custom backdoor

Analysed 2026-11-06 by Hermes

Open this case in the tool →

What happened

Cyberespionage data theft via spearphishing and custom backdoor

Dual vulnerability type

Expertise to Execution

Evidence coverage: 5/6 stages, weight 22/60

Decisive stage: containment (clearest evidence) · Alternative: Strategy to Leadership

Why this classification

(1) A government ministry of a state can be expected to protect its network with controls its own specialists know - filtering and verification of attachment/executable delivery, privilege tiering and directory hygiene around domain controllers, monitoring for C2 beaconing, egress visibility; Guyana's NDMA publicly claims exactly these arrangements (24/7 security operations centre, national CIRT, awareness programmes). (2) In practice, ESET's forensic report documents that targeted spearphishing emails referencing Guyanese public affairs led users to run double-extension executables downloaded from a compromised Vietnamese government website; the operators moved laterally and breached the target's internal network, created a new domain account and promoted it to the 'domain admins' group, dumped stored passwords from ntds.dit with ntdsutil.exe, and exfiltrated files, screenshots and clipboard data over a TEA-encrypted C2 channel; the intrusion was detected by ESET in February 2023, and Guyana's own cyber authority deployed investigators only after ESET's 5 October report, then declared there had been no successful infiltration. (3) The classification rests on the containment stage: the expectation that was not met sits with technical specialists (least-privilege tiering, controlled account creation, protection of ntds.dit - known remedies that need no leadership decision), and the chain broke in execution, because the documented path - arbitrary account creation, addition to domain admins, a domain-wide credential dump - shows those controls were never realised in operations; the entry gap points the same way (known delivery-stage remedies not executed), and no source documents a policy artefact, funding decision or ignored expert warning, so no Strategy-to-X or Expertise to Leadership variant is evidenced for the load-bearing gaps. The victim-side conditions that would let one code the human-factor marker at entry (training, filtering, verification procedures) are not documented and are deliberately not coded; recovery is not in play. The response stage shows a secondary Strategy to Leadership pattern - leadership's public narrative (no successful intrusion; the attempt was intercepted and its effects nullified) diverges from the primary forensic record, and disclosure came from the outside vendor, not the victim - but it followed the breach and made neither the intrusion possible nor its impact disproportionate, so it is recorded as contributing rather than decisive. RM41 (Expertise to Execution lens): the post-compromise escalation was a walk-through for specialists' worst-known path - a newly created account was added to the domain admins group and ntds.dit was dumped with ntdsutil - because privilege tiering and directory hygiene were never implemented in the network (causal role: Amplifying). RM30 (Expertise to Execution lens): compromise of the first couple of machines was not contained: operators moved laterally through the internal network and redeployed backdoors there, so the blast radius was the ministry network rather than the initial hosts (causal role: Amplifying). RM34 (Expertise to Execution lens): the SOC capability Guyana claims was not configured, staffed or tuned to catch this activity - ESET detected the campaign in February 2023 while the NDMA deployed its analysts only after the October 2023 publication (causal role: Obscuring). RM39 (Expertise to Execution lens): the inaccurate public disclosure had a material root - the authority had not established scope itself (it publicly awaited data from ESET and said its investigation was ongoing) and declared the attempt nullified before scoping was complete, so the public account contradicted the forensic record (causal role: Constraining).

Evidence by stage

Risk markers

Sources

Notes

Victim is unnamed in the EuRepoC row and in all sources; 'governmental entity in Guyana' (ESET) is the operative label - the NDMA statement confirms a spear-phishing attempt against a Government Ministry. Source conflict documented: ESET's forensic report (the row's attribution source, medium-confidence attribution to a China-aligned group) records a successful intrusion with lateral movement, domain-admin escalation, credential dumping and exfiltration over C2, while the NDMA - the victim's own cyber authority, not an independent investigator - stated on 10-11 October 2023 that no successful malware was found and that ESET exaggerated the threat; its investigation findings were never published and began months after the intrusion, during which DinodasRAT is capable of self-deletion. The incident was classified as the row describes it, per ESET's primary forensics; the NDMA denial does not trigger the 'No DV evidenced' regulator rule because it is not a regulator or formal independent finding and the primary forensic source documents specific gaps. ESET did not disclose what data was taken (The Record); EuRepoC scores the exfiltration as minor, non-sensitive. Attribution context (ESET): Korplug/PlugX use, the SOCU arrests of three persons in a money-laundering case involving Chinese companies in February 2023 disputed by the Chinese embassy, and Belt and Road economic interests - context for attribution only, not victim-side evidence. A separate victim of the same operators is out of scope here: the payload was hosted on a compromised Vietnamese government website (fta.moit.gov[.]vn), which ESET reported to VNCERT. Local Guyanese press (Kaieteur News reports of 10-11 October 2023 on a government probe) was blocked to automated access (bot wall) and is not cited; its substance is covered by the official DPI statement cited here.