2022 · O - Public administration and defence · Cyberespionage campaign via compromise of public-facing government servers and email accounts
Analysed 2026-11-06 by Hermes
Open this case in the tool →Cyberespionage campaign via compromise of public-facing government servers and email accounts
Expertise to Execution
Evidence coverage: 4/6 stages, weight 25/60
Decisive stage: entry (clearest evidence) · Alternative: Strategy to Execution
(1) The victims are government institutions - 48 government organisations compromised (10 of them foreign-affairs ministries) across 23 countries - which stand under standing sector expectations, later restated by the IMDA advisory, to keep internet-facing services patched, protect mail access with MFA and logon-attempt limits, segment public-facing servers from internal networks and monitor for post-exploitation activity. (2) In practice, from at least early 2022 until the vendor disclosure in March 2024, Earth Krahang repeatedly entered public-facing government servers through two known, already-patched vulnerabilities (CVE-2023-32315, CVE-2022-21587) or spear-phishing, guessed common passwords on OWA/ActiveSync mail portals, installed SoftEther VPN on the compromised public-facing servers to connect into private networks, dumped credentials, moved laterally, and packaged and exfiltrated mailboxes - while Trend Micro, not any victim, tracked the campaign from early 2022. (3) The classification rests on the entry stage: the technical remedies (the patches, MFA and logon-attempt limits prescribed by a regulator for exactly these techniques) existed and were knowable, yet the victim-side condition documented in the sources is that they were not realised in operations - servers stayed exploitable and mail accounts stayed guessable, and no source shows any leadership decision to defer or accept these controls, so the gap is material origin with an execution break, not a doctrinal one. The containment, detection and egress gaps corroborate the same pattern but rest on inference and cannot win a tie-break; the entry vector is load-bearing here not because the attacker used it but because the victim-side remediation and authentication state is documented. (4) RM18 (Expertise to Execution lens): both exploited vulnerabilities were known and remediable long before the observed exploitation, so patch execution on public-facing government servers lagged until it became the actors' standing foothold (causal role: Enabling). RM22 (Expertise to Execution lens): mail authentication was never brought to the prescribed standard - common-password brute-forcing of OWA/ActiveSync and a stolen authenticated Zimbra cookie gave legitimate-user access used for exfiltration (causal role: Enabling). RM30 (Expertise to Execution lens): one compromised public-facing server bridged into the private network via an installed VPN server, with RDP, WMIC and credential dumping widening the reach (in one country, 11 ministries), so containment did not bound the blast radius (causal role: Amplifying). RM34 (Expertise to Execution lens): detection capability was evidently not configured or maintained to surface LOLBins, LSASS access and protocol tunnelling - an external vendor monitored the campaign for roughly two years without any victim-side alert (causal role: Obscuring).
RM18 — Vulnerability / patch latencyRM22 — Authentication / credential exposureRM30 — Segmentation / containment weaknessRM34 — Monitoring / detection / alert-handling gapDiffuse multi-organisation campaign (70 victims in 23 countries per Trend Micro; the dataset rows list 17 receiver countries), so Country is left null per the method spec (multi-organisation campaigns are never guessed). Sector coded O because the documented majority of victims are government organisations (48 compromised government organisations incl. 10 foreign-affairs ministries); the output check requires a single CBS SBI entry, though some victims sit in H/Q/J/K/P/S. Attribution to I-Soon is Trend Micro's surmise from Earth Lusca overlaps, not confirmed; Earth Krahang is assessed as a separate intrusion set. The case-file source_url list contains one link unrelated to this incident (a Wired article on an Apple M-chip flaw); it was not used or cited. Victim-side origin/break answers for the containment, detection and egress stages are coder inferences - no source documents decision-making at the unnamed victim population - which is why only the entry stage is primary-sourced and the classification claim is an inference. The row description matches the Trend Micro report; no victim/description mismatch found. The victim population is established as a category (EuRepoC names only receiver categories), which is the basis for the victim-established claim.