1. Executive summary
Multiple "confused deputy" vulnerabilities persist in Google Cloud Platform (GCP) and Microsoft Azure, enabling attackers to acquire administrative-level permissions and bypass cloud provider access controls. The flaw class abuses intermediary services that possess elevated authority, tricking them into performing privileged actions on behalf of an unauthorised caller. For EMEA financial services, exploitation grants direct administrative control over cloud infrastructure, negating IAM boundaries and threatening the integrity and availability of regulated workloads. No verified CVE data, CVSS scores, or CISA-KEV exploitation states were resolved for this item; treat the specific vulnerability scope as unconfirmed pending vendor advisories.
2. Regulatory framing
No specific DORA/NIS2 article is directly engaged by this item. While the vulnerabilities affect cloud infrastructure, the provided source material does not contain distinctive facts — such as a confirmed major incident at a client or a specific third-party provider failure — required to trigger the narrow obligations under DORA Arts. 17–19, 24, 28–30, NIS2 Arts. 21(2)(d) or 23, or UK NIS 2018. Generic cloud security hygiene does not meet the threshold for these articles.
3. Technical analysis & attack chain
The source material describes a class of "confused deputy" vulnerabilities affecting both Google Cloud and Microsoft Azure. A confused deputy is a privilege escalation flaw where a privileged intermediary service (the "deputy") is tricked by a lower-privileged attacker into misusing its authority to access or modify resources the attacker cannot directly touch. In cloud environments, this typically manifests when a service with broad IAM permissions accepts requests or tokens without sufficiently validating the calling principal's authorisation for the target action.
The BSI advisory (WID-SEC-2026-2321) corroborates that multiple vulnerabilities exist across various Microsoft Azure components, enabling privilege escalation and denial of service. The BSI advisory does not specify the affected components, CVE identifiers, or technical exploitation mechanics.
Attack chain (reconstructed from the vulnerability class; specific steps are unconfirmed for these instances)
- Initial position: Attacker holds low-privilege access to the cloud tenant (e.g., a standard user account, a misconfigured service principal, or a guest identity).
- Deputy identification: Attacker identifies a cloud service or API endpoint that operates with elevated permissions (the confused deputy).
- Request forgery: Attacker sends a crafted request to the deputy service, specifying a target resource the attacker cannot directly access but the deputy can.
- Privilege misuse: The deputy service, failing to verify the caller's authority over the specific target resource, performs the action using its own elevated permissions.
- Outcome: Attainer achieves administrative-level permissions, bypassing cloud provider access controls. In Azure, this may additionally enable denial of service against dependent services.
Confidence caveat: The specific vulnerable components, CVE identifiers, exploit payloads, and observed threat actor activity are not present in the provided sources. The attack chain above is derived from the general "confused deputy" vulnerability class described in the article and is consistent with the BSI's confirmation of Azure privilege escalation flaws. No named threat actor is identified; no MITRE ATT&CK profile is available. Single-sourced claims (Dark Reading) are corroborated by the BSI advisory for Azure but not for GCP.
4. Mitigation & containment
P1 — Within 24 hours
- Review cloud tenant audit logs (GCP Cloud Audit Logs, Azure Activity Log) for anomalous privilege escalation events: unexpected role assignments, service principal credential creation, or API calls from low-privilege identities targeting administrative resources.
- Enforce least-privilege IAM: audit custom roles and service principal assignments for overly broad permissions (e.g.,
Owner,Contributorin Azure;roles/owner,roles/editorin GCP). Remove unused permissions. - Enable and review Conditional Access (Azure) and Context-Aware Access (GCP) policies to restrict privileged API calls to trusted locations and devices.
P2 — Within 72 hours
- Monitor for confused deputy patterns: alert on scenarios where a low-privilege principal triggers a high-privilege service action. In Azure, alert on
Microsoft.Authorization/roleAssignments/writeevents initiated by non-privileged service principals. In GCP, alert onsetIamPolicycalls from identities lackingresourcemanager.projects.setIamPolicy. - Apply any vendor-published patches or configuration guidance. The BSI advisory references Azure component vulnerabilities; monitor the Microsoft Security Response Center (MSRC) and BSI advisory WID-SEC-2026-2321 for specific KB articles or configuration changes. No specific patch identifiers are available in the source material.
P3 — Within 7 days
- Conduct a targeted IAM audit: map all service-to-service trust relationships and identify deputies (services with broad permissions that accept requests from lower-trust callers). Document and restrict the delegation chain.
- Validate that break-glass accounts and privileged access workflows are not exposed to confused deputy abuse paths.
- Engage cloud provider support for confirmation of affected services and remediation timelines if vendor advisories remain unpublished.
5. Indicators of compromise
No indicators of compromise available in the source material.
Behavioural indicators
| Behaviour | Where to observe | Confidence |
|---|---|---|
| Low-privilege identity performing role assignment or IAM policy modification | Azure Activity Log; GCP Cloud Audit Logs (Admin Activity) | Medium — derived from vulnerability class, not confirmed observation |
| Service principal with no prior administrative activity suddenly escalating privileges | Azure AD sign-in logs and Activity Log; GCP Cloud Audit Logs | Medium — derived from vulnerability class |
| Unexpected denial-of-service patterns targeting Azure components | Azure Service Health; Azure Monitor | Low — BSI confirms DoS vector exists; no specific pattern described |
6. Detection
Insufficient indicators to author detection rules. The source material does not contain specific artefacts (strings, file paths, command-line arguments, registry keys, mutex names, or network signatures) associated with exploitation of these vulnerabilities. The behavioural indicators in §5 should be implemented as log-based alerting rules using the described patterns.
7. Sources
- Dark Reading — "'Confused Deputy' Flaws Persist in Google Cloud, Microsoft Azure" — https://www.darkreading.com/cloud-security/confused-deputy-flaws-google-cloud-microsoft-azure — 2026-07-27
- BSI Germany (CERT-Bund) — "[NEU] [hoch] Microsoft Azure: Mehrere Schwachstellen" (WID-SEC-2026-2321) — https://wid.cert-bund.de/portal/wid/securityadvisory?name=WID-SEC-2026-2321 — accessed 2026-07-28
8. Adverse Trace position
Severity cannot be definitively assessed because no verified CVSS scores or CISA-KEV exploitation states were resolved for this item. However, the vulnerability class — confused deputy enabling administrative privilege escalation and access control bypass in major cloud platforms — is inherently high-impact for EMEA financial services running regulated workloads on GCP or Azure. The BSI's independent confirmation of multiple Azure privilege escalation and DoS vulnerabilities (rated "hoch") corroborates the Dark Reading reporting for Azure; the GCP claim remains single-sourced. We assess that clients with cloud-hosted workloads should treat this as a high-priority IAM hygiene and monitoring issue even in the absence of specific CVEs. Adverse Trace will update this advisory when vendor advisories publish specific CVE identifiers, affected component lists, and remediation guidance.
Published via PulseTrace — Adverse Trace threat intelligence.