Blog
AI-Powered Teams Spear-Vishing Meets a Microsoft Teams Zero-Day
The attack against the financial-sector organization had two critical elements.
First, attackers used an AI-backed Microsoft Teams spear-vishing attack. Employees received Teams calls from an external account impersonating an internal developer they knew. The caller also sounded like the legitimate employee, making the interaction appear credible via AI cloned voice. The attacker then persuaded users to share their screens and grant remote control through Teams, effectively turning a trusted collaboration platform into an initial-access channel.
The Second Problem: A Microsoft Teams Zero-Day
The incident becomes even more concerning when combined with a reported Microsoft Teams zero-day affecting remote-control containment.
A zero-day generally refers to a previously unknown or unaddressed security weakness for which defenders may not yet have an effective fix or established mitigation. In this case, the concern relates specifically to an already-established Teams remote-control session.
Testing described in the security advisory report shared on GitHub in June 2026 showed that when the controlling account’s Microsoft Entra ID sessions were revoked, the account was forced to authenticate again and could no longer initiate new authenticated interactions. However, the remote-control session that had already been established reportedly remained active, allowing control of the remote computer to continue.
This creates an important incident-response problem:
Revoking the attacker’s identity session may not necessarily terminate the attacker’s existing Teams remote-control session.
Microsoft reviewed the reported behavior through MSRC but determined that it did not meet Microsoft’s criteria for a vulnerability requiring servicing. Therefore, this should not be presented as a Microsoft-confirmed zero-day or authentication bypass. From a defensive perspective, however, the reported behavior represents a significant potential containment gap that security teams need to consider.
Why the Combination Is So Dangerous
The two issues complement each other:
AI voice impersonation → trusted Teams call → screen sharing → remote control → interactive access → Teams zero-day containment problem
The AI-backed spear-vishing attack gets the attacker onto the workstation, while the reported Microsoft Teams zero-day may complicate the SOC’s ability to terminate that access once it has already been established.
That is the real security story.
Organizations can no longer assume that a familiar name or even a familiar voice proves who is behind a Teams call. And during incident response, disabling an account or revoking sessions should not automatically be considered proof that an active remote-control connection has been terminated.
Contact
Want to secure your company?
Talk to us.
You might also like this
- Blog
- Blog
- Blog