← Breach analysis by the Dual Vulnerabilities framework

Unspecified multi-sector targets data theft and ransomware wave (UNC3944/Scattered Spider, 2023)

2023 · J - Information and communication · Data theft and ransomware deployment via social engineering (multi-sector campaign)

Analysed 2026-11-07 by Hermes

Open this case in the tool →

What happened

Data theft and ransomware deployment via social engineering (multi-sector campaign)

Dual vulnerability type

Strategy to Execution

Evidence coverage: 3/6 stages, weight 15/60

Decisive stage: containment (clearest evidence) · Alternative: Expertise to Execution

Why this classification

(1) The wave targeted telecommunications companies, business process outsourcers and hospitality, retail, media and financial-services organisations, where the standing expectation is that a mandated endpoint-security and logging baseline applies to every system in the estate and that account-recovery processes verify an employee's identity independently of information an impersonator can collect. (2) Mandiant's report documents intrusions that entered through SMS phishing and service-desk impersonation - callers supplied usernames, employee IDs and PII sufficient to pass verification and received MFA codes, password and MFA-token resets - and then worked inside the victims' cloud: actors created Azure VMs with public IPs that did not have the organization's mandated security and logging software installed, giving unmonitored access to trusted systems; added rogue federated identity providers to Entra for golden SAML logins; exported credentials from the victims' own HashiCorp Vault and CyberArk deployments; and in the incident behind this row's AWS S3 detail retrieved utilities from a victim-owned S3 bucket via an Intune PowerShell orchestration that disabled firewall rules and Defender protections before executing the ALPHV payload. (3) The classification rests on the containment stage and it is a diagonal gap: the formal artefact existed - the report itself references the victims' mandated security and logging software - yet the architecture did not enforce it, so attacker-built systems ran unmonitored inside the estate and the victims' own management orchestration carried the ransomware payload. That is a knowing-doing gap between a mandated baseline and realised operations, not generic weak security; the entry mechanism is the wave's best-known element but its material origin is only inferred (no source records any coded victim's authentication policy), so entry is not decisive, and egress, recovery and response yield no documented victim-side gap for an identified victim. RM23 (Strategy to Execution lens): management-tool security, approval and deployment safeguards were not in place, so the victims' own Intune PowerShell orchestration was used from inside the environment to fetch and run the ALPHV payload after disabling firewall and Defender protections (causal role: Amplifying). RM45 (Strategy to Execution lens): independent verification was never built into the service-desk account-recovery process, letting impersonators pass knowledge-based checks with openly available PII and obtain MFA codes, password and MFA-token resets (causal role: Enabling). RM34 (Strategy to Execution lens): monitoring requirements were not operated as intended against cloud control-plane activity - the actors' persistence is described as hard to monitor for and detect and they used residential proxies to fly under the radar of security monitoring tools, so intrusions ran for days (causal role: Obscuring). RM13 (Strategy to Execution lens): supplier-security requirements did not translate into controlled third-party operation - in one intrusion the victim's help desk gave an MFA-token reset for credentials exposed on a RECORDSTEALER-compromised laptop of an IT outsourcing company contracted by the victim (causal role: Enabling).

Evidence by stage

Risk markers

Sources

Notes

Aggregated campaign row: EuRepoC incident 2624 codes the mid-2023 UNC3944 wave as reported by Mandiant on 14-09-2023; no single victim organisation is named. All victim-side evidence is the group-wide TTP pattern documented in the vendor report across its observed intrusions, not incident-confirmed for an identified victim, and the victim itself is not established - hence the pattern source types and the confidence cap. The live mandiant.com URL delivered a JavaScript shell without the article body in this environment, so the report was read and cited via the opened Wayback snapshot of 15-09-2023. The row's sentence about downloading malicious files from a victim's AWS S3 and executing ALPHV is a compressed paraphrase of Mandiant's Intune/victim-owned-S3-bucket paragraph. Country is null: the case file carries no receiver_country and the campaign spans several countries. The receiver categories are mixed (corporate, media, telecommunications, finance); the method spec would leave Sector blank for a diffuse campaign, but the output check requires one CBS SBI entry, so J Information and communication is recorded as the wave's primary coded sector (Mandiant: targeting began at telecommunications and business process outsourcers). Disclosure came from the IT-security company, not from any victim.