# The Pillar Cracks Three Ways: Azure Identity Infrastructure Takes Three Max-Severity Hits in One Patch Tuesday

> Source: <https://forkast.news/the-pillar-cracks-three-ways-azure-identity-infrastructure-takes-three-max-severity-hits-in-one-patch-tuesday/>
> Published: 2026-09-17 16:35:06+00:00

## Azure Identity Infrastructure Vulnerability Cluster

The September 2026 Patch Tuesday cycle included 964 total CVEs, the largest volume recorded to date. Within this update, three high-severity vulnerabilities targeted the Azure identity stack. These flaws, all resolved through server-side remediation, underscore the concentration of risk within core authentication and authorization services.

## Technical Breakdown of CVEs

CVE-2026-83711 affects Azure AD B2C. It carries a CVSS score of 10.0 and is classified as CWE-639, an authorization bypass through user-controlled keys. The vulnerability allows unauthenticated, network-based access with no user interaction required (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:N). Microsoft published this disclosure out-of-band on September 3, 2026.

CVE-2026-70352 impacts Azure AI Language. It is rated at 10.0 and categorized as CWE-306, missing authentication for critical functions. Like the B2C flaw, it permits unauthenticated network access (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:C/C:H/I:H/A:H). This was also disclosed out-of-band on September 3, 2026.

CVE-2026-83941 affects Entra ID. Microsoft assigns a CVSS score of 9.9 for this elevation of privilege flaw, which involves missing authorization. However, [Tenable and NVD](https://www.tenable.com/cve/CVE-2026-83941) report a score of 8.8 (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H). This discrepancy highlights variations in risk assessment methodologies across the security industry.

## Server-Side Remediation and Trust

All three vulnerabilities were addressed by Microsoft within its own infrastructure. No customer-installed patches were required. This remediation model shifts the burden of discovery and resolution entirely to the service provider. While this removes the immediate operational task of patching, it centralizes the security dependency on the provider’s internal processes.

## Azure AD B2C Lifecycle Status

The presence of a 10.0-rated vulnerability in Azure AD B2C occurs while the platform is in maintenance mode. The service reached its end-of-sale date in May 2025, though support is guaranteed until at least May 2030. Microsoft currently directs users toward migration to Entra External ID.

## Identity Risk in AI Infrastructure

The inclusion of Azure AI Language in this vulnerability cluster indicates that identity-based attack surfaces now extend into AI service layers. The failure to enforce authentication in this instance mirrors the risks found in traditional identity systems. Securing AI endpoints requires the same authentication controls applied to core identity infrastructure.

## Structural Patterns in Identity Security

These vulnerabilities reflect recurring failures in authorization and authentication logic within identity and privileged access management platforms. Similar structural defects have been documented in the [Azure SRE Agent](https://forkast.news/cvss-9-9-flaw-in-azure-sre-agent-breaks-obo-flow-extending-blast-radius-beyond-the-agent/) (CVE-2026-62830), [Delinea Secret Server](https://forkast.news/two-unauthenticated-lockpicks-for-the-enterprise-vault-delinea-secret-servers-double-critical-disclosure/), and [Cisco ISE](https://forkast.news/the-gatekeeper-is-the-door-cisco-ises-nine-cve-disclosure-and-the-identity-infrastructure-attack-surface/). System complexity continues to correlate with high-severity disclosures in these environments.

## What to Watch

- **B2C Migration:** The vulnerability in a maintenance-mode platform raises questions about the security posture of legacy identity services approaching end of support.
- **Scoring Discrepancy:** The variance between Microsoft’s 9.9 and NVD’s 8.8 for CVE-2026-83941 reflects ongoing divergence in cloud-native vulnerability assessment.
- **AI Identity Surface:** Azure AI Language’s inclusion in this cluster establishes AI service endpoints as identity-layer targets, not just model-layer targets.
