Introduction
A critical flaw in Microsoft Authenticator lets an unauthorized attacker silently steal a user's work account access token over a network. The token grants full access to whatever enterprise applications and data that user is authorized to reach. CVE-2026-41615 was disclosed on May 14, 2026, with a CVSS base score of 9.6. It turns the tool organizations rely on for multifactor authentication into a way to steal tokens, which effectively bypasses the authentication layer the app was designed to enforce.
Technical Information
CVE-2026-41615 is classified under CWE-200: Exposure of Sensitive Information to an Unauthorized Actor. Microsoft Authenticator can be manipulated into obtaining an access token and transmitting it to an attacker-controlled location, without clearly telling the user what specific access is being granted.
Attack Flow
The exploitation chain has five steps:
- Initial contact: The attacker crafts a malicious request that looks like a legitimate authentication prompt and delivers it to the target user over the network.
- User approval: The victim believes the request is genuine and approves the prompt on their device.
- Token acquisition: After approval, the Authenticator app obtains an access token for the user's work account on the user's behalf.
- Token exfiltration: The app silently transmits the token to a location the attacker controls. At no point is the user clearly told what specific access is being granted.
- Post-exploitation: The attacker uses the stolen token to reach data and services the user is authorized to use, including enterprise applications and data repositories.
CVSS Breakdown
Several factors compound to produce the 9.6 score:
| Metric | Value | Implication |
|---|---|---|
| Attack Vector | Network | Remote exploitation, with no physical or local access needed |
| Privileges Required | None | No prior authorization or access required to initiate the attack |
| User Interaction | Required | The victim must approve a prompt. The exploit cannot be fully automated |
| Scope | Changed | Compromise extends beyond Authenticator to external enterprise resources |
| Confidentiality | High | Complete disclosure of the access token and associated data |
| Integrity | High | Attacker can modify data within authorized services |
| Availability | High | Attacker could disrupt services accessible via the stolen token |
The Scope Change rating is what lifts this from a high-severity issue to a critical one. Microsoft explains that an exploited vulnerability can affect resources beyond the security scope the vulnerable component manages. Authenticator is the vulnerable application, but the stolen token compromises external enterprise applications and data repositories that entirely different security authorities manage.
What We Do Not Know
The current vendor advisory does not say which exact versions of Microsoft Authenticator are vulnerable. It also details remediation for Android devices explicitly, but does not confirm whether iOS versions of the app are equally affected or need similar updates.
Affected Systems and Versions
The vulnerability affects Microsoft Authenticator on Android. The vendor advisory gives no exact version numbers or version ranges for the vulnerable builds, and Microsoft has not yet confirmed or denied iOS impact. Until Microsoft provides more granular version guidance, organizations running Microsoft Authenticator on Android devices should treat all unpatched versions as potentially vulnerable.
Mitigation
Microsoft has released an official fix. Remediation by platform:
| Platform | Update Mechanism | Required Action |
|---|---|---|
| Android | Automatic Updates | Users with automatic updates enabled will receive the patch without manual intervention |
| Android | Manual Updates | Users with automatic updates disabled must manually download the latest version via the Google Play Store |
| iOS | Unspecified | Monitor vendor channels for clarification on iOS impact and update requirements |
Compensating Controls
Patching is the primary defense, but organizations should layer other protections on top of it:
- Microsoft Entra ID Protection can detect, investigate, and remediate identity-based risks. Security teams should enable these features and monitor them for anomalous sign-in activity.
- Conditional access policies should be enforced so they can detect potential misuse of stolen tokens, such as sign-ins from unexpected locations or devices.
- User awareness training applies directly, because the exploit chain requires the victim to approve a malicious authentication prompt. Users who know to deny unexpected or suspicious prompts make successful exploitation less likely.
Threat Intelligence
As of the May 14, 2026 disclosure, the Exploit Code Maturity for CVE-2026-41615 is rated Unproven. Microsoft reports no publicly available exploit code, and any potential exploit remains theoretical for now. There is no evidence of active exploitation in the wild, and no specific threat actors have been linked to this vulnerability.
Enterprise access tokens are still high-value targets for advanced persistent threats and initial access brokers. A network-based attack vector, no required privileges, and access to enterprise resources together make this vulnerability an attractive candidate for weaponization. Organizations should use the current pre-exploit window to deploy the fix before functional exploit chains emerge.
