A critical security flaw in Microsoft Entra ID (formerly Azure AD) involved the use of undocumented 'actor tokens' and a vulnerability in the Azure AD Graph API (CVE-2025-55241), which could allow attackers to hijack any company's tenant and access sensitive data.

The flaw originated from improper validation of these actor tokens via the legacy Azure AD Graph API, enabling impersonation of any user, including Global Admins, across most tenants.

The security researcher uncovered methods to brute-force user netIds and exploit B2B guest trust relationships, potentially propagating tenant breaches extensively.

While telemetry for token requests is limited, modifications to the API do generate audit logs, which can help detect misuse if suspicious activity is observed.

Attackers could craft impersonation tokens with known tenant and user IDs to query or modify tenant resources via the Azure AD Graph API, which does not generate sufficient logs to detect such abuse.

Researcher Dirk-jan Mollema demonstrated that exploiting actor tokens could grant Global Admin privileges in any Entra ID tenant, with all related actions remaining unlogged.

By manipulating tenant IDs and user netIds, attackers could access data, modify configurations, and escalate privileges without detection or alerts.

Actor tokens are JWTs issued by a legacy service, unsigned, with a 24-hour validity, and can impersonate any user within a tenant, posing a major security risk due to their lack of logging, revocation, and conditional access controls.

These tokens, used in backend service-to-service communication, lack proper security controls, enabling undetectable impersonation and tenant compromise.

Microsoft responded swiftly with patches and mitigations, including blocking actor token requests for the Azure AD Graph API, and issued CVE-2025-55241 to address the vulnerability.

The researcher emphasized the severity of this vulnerability, noting it could lead to widespread tenant breaches and data exfiltration, prompting recent mitigations and ongoing vigilance.

Microsoft confirmed that a patch was deployed on September 4, 2025, to resolve the issue and is phasing out actor tokens due to their security flaws, including lack of logging, revocation, and bypassing Conditional Access.