← Breach analysis by the Dual Vulnerabilities framework

Slovak Government Spear-phishing Espionage (APT29/SVR, 2021)

2021 · Slovakia · O - Public administration and defence · State-sponsored spear-phishing espionage campaign (backdoor implantation, no misuse recorded)

Analysed 2026-11-06 by Hermes

Open this case in the tool →

What happened

State-sponsored spear-phishing espionage campaign (backdoor implantation, no misuse recorded)

Dual vulnerability type

Expertise to Execution

Evidence coverage: 2/6 stages, weight 16/60

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

Why this classification

(1) As a EU/NATO member state running ministries and a diplomatic service under national cyber-security arrangements, the expectation at the victim was that persistent adversary presence on government networks - a backdoor beaconing over HTTPS to public internet hosts - would be found by its own monitoring. (2) In practice, the SVR-linked APT29 group ran spear-phishing campaigns against Slovak officials and diplomats from February to July 2021 using ISO disk images disguised as communications of the country's own National Security Authority, installing Cobalt Strike; the campaign was discovered only in March 2021 by the private firm IstroSec working from public sources and community threat intelligence, C2 servers were still active in June, and the case became public in August 2021 through ESET and IstroSec rather than through the victim. (3) This is a knowing-doing gap inside the material column, not merely weak security: the expertise required to detect the intrusion demonstrably existed and was cheap to exercise - a small external firm reconstructed the whole campaign and its infrastructure from publicly available information alone - yet the targeted administration's own detection capability did not turn that available knowledge into an intervention for months. The classification rests on the detection stage, documented by a primary forensic report; the entry gap (a publicly documented phishing chain that filtering and application control did not stop) points to the same type but rests partly on coder inference, so detection is the decisive stage by clearest evidence, and no competing stage was found. (4) RM34 (Expertise to Execution lens): the victim's monitoring and alert handling were not configured, staffed or tuned to spot months of Cobalt Strike beaconing that external researchers found from public data alone, leaving the intrusion hidden (causal role: Obscuring). RM24 (Expertise to Execution lens): verification and awareness controls did not hold against lures abusing the trusted name of the national security authority - successive campaigns over five months kept succeeding against diplomatic staff (causal role: Enabling). RM28 (Expertise to Execution lens): mail filtering let ISO/IMG attachments posing as Word documents reach targets and endpoints executed the LNK-to-DLL chain unimpeded, although the technique was already publicly reported and detectable (causal role: Enabling).

Evidence by stage

Risk markers

Sources

Notes

EuRepoC row; victim detail is thin: no official Slovak statement on the incident was found, so the disclosure and response picture rests on the security firms' reporting (case field source_disclosure: incident disclosed by IT-security company). Attribution to APT29/SVR is a technical attribution by IstroSec, ESET and Microsoft based on TTPs and infrastructure, not an official state designation of this specific campaign; the ESET Twitter post cited in the case file could not be opened directly and ESET's assessment is taken via The Record article. The dataset's multi-valued receiver_country (Slovakia; Czech Republic; Europe) reflects the related campaigns IstroSec and ESET describe on the same infrastructure, not separate incidents. The dataset records the incident as hijacking without misuse with no data theft, so the egress stage is treated as not in play. Malware sample, VirusTotal/Any.Run and IOC links referenced by IstroSec were deliberately not opened (research boundaries). No action requiring consent was needed in this run.