2015 · Germany · O - Public administration and defence · State-sponsored network intrusion with mass data theft
Analysed 2026-11-06 by Hermes
Open this case in the tool →State-sponsored network intrusion with mass data theft
Expertise to Execution
Evidence coverage: 6/6 stages, weight 36/60
Decisive stage: containment (clearest evidence) · Alternative: Expertise to Leadership
(1) As the German federal parliament whose network (Parlakom, more than 20,000 computers for MPs, factions and administration) handles legislative and government-relevant material, the Bundestag was expected to have the specialist-level protections of its trade realised: mail and endpoint defences against standard malware, tiered privileged access and directory protection, and monitoring of bulk data leaving the network. That the knowledge existed in the organisation is shown by the specially secured compartments (NSA committee, Geheimschutzstelle, personnel administration), which stayed out of reach precisely because they ran on separate networks. (2) In practice the entry was ordinary - one phished Die Linke MP's computer, simple unobfuscated malware - but from there the attackers moved through a flat network with administrative reach: they harvested the credentials of five of six domain-admin accounts with mimikatz, took over the directory service, and exfiltrated at least 20 GB of archives and documents unnoticed. The Bundestag admins' own first suspicion of 13 April 2015 took until 12 May to escalate, and only because an English company found the stolen documents on its own servers; BSI experts declared the network indefensible and it had to be abandoned and rebuilt. The entry vector is not load-bearing; what made the impact disproportionate was the flat directory architecture and the missing credential, containment and egress controls. (3) This is a knowing-doing gap within the material column, not a strategy that failed to reach operations: every unmet expectation was a known technical remedy held by specialists (the BSI itself noted the attackers used methods and public tools typical of less professional actors), and the sources show the architecture simply did not realise them - reusable domain-admin credentials on ordinary systems, no containment between faction and administration reach, no egress detection. No source shows leadership deciding, funding or deferring these specific controls, so the break is execution, not leadership. A second pattern appears in the response stage: the administration withheld BSI's classified analysis that the network was indefensible from MPs, who learned details from the press - an Expertise to Leadership reading that is contributing, not decisive. (4) RM41 (Expertise to Execution lens): from one infected MP's PC the attacker reached domain-wide control by harvesting reusable credentials of five of six domain-admin accounts with mimikatz, because expert-level privilege tiering and directory hardening were never realised in the operational architecture, scaling the blast radius to the whole directory (causal role: Amplifying). RM30 (Expertise to Execution lens): containment capability was materially insufficient - one heavily branched flat zone the attackers could roam freely, which BSI experts judged beyond repair and recommended abandoning, while only specially separated networks stayed safe (causal role: Amplifying). RM34 (Expertise to Execution lens): the alert was raised by the Bundestag admins' own first suspicion on 13 April but not acted upon for nearly a month, the warning arriving only via an external tip - operational alert handling did not follow through on what had been flagged (causal role: Obscuring). RM35 (Expertise to Execution lens): overt staging scripts and a clear-text tunnel beaconed gigabytes out without the victim noticing at all, since an outside company's discovery first surfaced the theft - egress visibility that a BSI-monitored parliament network should have had was never operationalised (causal role: Amplifying).
RM41 — Privilege-escalation / directory-hygiene pathRM30 — Segmentation / containment weaknessRM34 — Monitoring / detection / alert-handling gapRM35 — Egress / exfiltration monitoring gapSource caveats: the dataset's source_url mixes in articles about other incidents (a 2023 APT28 campaign against Ukrainian government mail servers, a La Stampa drone report, CDU-hacking articles); only the articles listed under Manual source-article were opened and are specific to this incident. The entry vector is not fully settled (per the internal IuK/BSI documents: malicious attachment or drive-by download; the 2020 prosecution reporting via ZDNet describes a UN-themed boobytrapped document) and is not load-bearing for the classification. Guarnieri's forensic report was commissioned by the Die Linke faction and covers the faction servers; the Bundestag administration itself never confirmed the APT28 reading - public attribution rests on the BfV statement of May 2016 (BBC), the Federal Prosecutor's 2020 arrest warrant against Dmitry Badin of GRU unit 26165 (ZDNet) and the October 2020 EU sanctions named in the case description. The 12 May 2015 exfiltration warning came via a foreign company's discovery - third-party reporting, corroborated across heise's coverage of the leaked IuK protocols. Related rows checked: row 3625 (CyberBerkut's claimed January 2015 DDoS on chancellor.de/bundestag.de) and row 2287 (2022 e-mail theft from German IT companies) mention the Bundestag context but are different incidents, so no recurring-weakness coding.