← Breach analysis by the Dual Vulnerabilities framework

Unidentified victim EarlyRAT infection (Andariel, 2023)

2023 · S - Other service activities · State-sponsored malware infection (new RAT) without misuse

Analysed 2026-11-07 by Hermes

Open this case in the tool →

What happened

State-sponsored malware infection (new RAT) without misuse

Dual vulnerability type

No DV evidenced

Alternative: none defensible

Why this classification

(1) Strategic expectation at the victim: nothing is documented - the victim is not named in any source (Kaspersky's Securelist report and press release, BleepingComputer, Security Affairs), never spoke about the incident, and the EuRepoC row carries no receiver name, country or sector, so no policy, standard or leadership statement of the victim is known. (2) What happened in practice: Kaspersky found a previously undocumented RAT (EarlyRAT) in one of its Log4j-exploitation cases - the campaign exploited Log4Shell on internet-exposed services in mid-2022 to breach corporate networks and then ran DTrack plus off-the-shelf tools - and on hunting for more samples identified phishing documents that dropped EarlyRAT via macros; the row's EarlyRAT instance appeared to have been deployed through a phishing campaign. The incident is recorded in EuRepoC as hijacking not used, with no disruption, no data breach or exfiltration and no economic impact, detected externally by Kaspersky during an unrelated investigation and disclosed by that company on 28 June 2023. (3) Why this is not classifiable: the reporting confirms the intrusion and its (null) impact only. The entry mechanism is documented, but it is the attacker's technique - reading a DV-Type from macro-enabled phishing documents would be the rejected attacker-technique reasoning; RM24's exclusion forbids coding a phishing success without evidence of the organisational condition (training, verification, procedure), RM18 does not apply to this victim because the EarlyRAT deployment was via phishing rather than an unpatched Log4j instance, and the detection by an outside vendor does not by itself evidence a monitoring gap at this particular victim (RM34's exclusion forbids inferring suppression from absent evidence). No source states any expectation or remedy the victim knew and did not act on, so neither the origin nor the break can be answered from evidence at any stage; per the decision rule, a case where reporting confirms impact only is Unknown, not a forced DV-Type. (4) Risk-marker lens sentences are skipped for Unknown.

Sources

Notes

Atypical, victim-anonymous case: every source (Kaspersky Securelist report and press release, BleepingComputer, Security Affairs) reports only the attacker's tradecraft; the victim's identity, sector and country are undisclosed, so Country and Year are null. The victim's economic activity is unknown in every source (the workbook's own receiver_category is not filled for this row, and the comparable Andariel/Log4Shell row in the same sheet codes it Unknown), so the sector could not be established; 'S Other service activities' is recorded only because the output check requires one exact CBS SBI entry to process the row - readers should treat the victim sector as unknown. The only dated point is Kaspersky's disclosure of 28 June 2023 (also the EuRepoC attribution date); Kaspersky and BleepingComputer place the related Log4Shell campaign in mid-2022, but this row's EarlyRAT infection was deployed via phishing and is not dated in any source, so the incident year is not established - the year in the breach name is the reporting year. The case file itself says the EarlyRAT instance 'appeared to have been deployed through a phishing campaign' - delivery method stated with reservation in the source. EuRepoC scores the incident as hijacking not used (intensity 1): no disruption, no data breach or exfiltration, no economic impact. The third source_url in the case file (Security Affairs newsletter round 426) was opened but carries no case-specific detail beyond a headline; it is not cited. No victim-side evidence exists to code risk markers, so none is recorded (permitted for Unknown).