CVE-2026-119990 – Azure OpenAI Elevation of Privilege Vulnerability
When a critical function forgets to ask who you are, anyone on the network could walk in with elevated privileges.
CVE-2026-119990 is a Critical elevation of privilege vulnerability in Azure OpenAI. A critical function was missing authentication, which could let an unauthenticated attacker elevate privileges over the network. The attack needs no credentials and no user interaction, which is why it carries a CVSS score of 9.8. Microsoft lists customer action as required for this issue.
CVSS Score: 9.8
SEVERITY: Critical
THREAT: A remote attacker with no credentials could reach an unprotected critical function in Azure OpenAI and gain elevated privileges, opening the door to data and resources tied to the service.
EXPLOITS: This vulnerability is not known to be exploited. It was not publicly disclosed before the fix, and no public proof-of-concept exploit code is identified.
TECHNICAL SUMMARY: The flaw is missing authentication for a critical function (CWE-306) in Azure OpenAI. A function that should have checked who was calling it did not, so an attacker could use it without proving their identity and end up with privileges they should never have. The CVSS v3.1 vector is AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H: network-based, low complexity, no privileges, and no user interaction. Successful exploitation has high impact on confidentiality, integrity, and availability. Microsoft does not give further technical detail on the affected function.
EXPLOITABILITY: The affected product is Azure OpenAI, and the fix is described in Microsoft's release notes. An attacker would send requests over the network to the unprotected function, with no credentials or user action needed. The specific affected versions or deployments are not listed in the data.
BUSINESS IMPACT: Organizations increasingly connect Azure OpenAI to internal data, applications, and workflows. Elevated access to that service could expose prompts, model outputs, connected data, and business logic, or let attackers misuse the service and run up costs. A breach here could leak sensitive information, damage trust in AI-driven processes, and create compliance problems.
WORKAROUND: No official mitigation or workaround is listed. Follow the actions in Microsoft's release notes for Azure OpenAI, and in the meantime restrict network access to Azure OpenAI resources with private endpoints or firewall rules, review access keys and role assignments, and watch activity logs for unusual requests.
URGENCY: A 9.8 CVSS score, Critical severity, and an attack that needs no credentials and no user interaction put this at the top of the list. Even though no exploitation is known, missing-authentication bugs are easy to find and abuse once details leak. Because Microsoft marks customer action as required, cloud and AI teams should confirm their Azure OpenAI resources are covered right away rather than assume the fix is automatic.
Key Details
- Attack Vector
- Network
- Attack Complexity
- Low
- Privileges Required
- None
- User Interaction
- None
- CWE Classification
- CWE-306