Skip to content

SecurityMicrosoft

Fixed

Microsoft fixes M365 Copilot prompt injection that let emails plant convincing phishing in Copilot output

Microsoft 365 Copilot at a glance

Microsoft published CVE-2026-26133 on March 12, 2026: an AI command injection in Microsoft 365 Copilot where a malicious email could make Copilot show authoritative-looking phishing messages. It is rated Important (CVSS 7.1), and Microsoft lists fixed builds of Copilot, Outlook, Teams, Office, and Edge apps.

Microsoft 365 Copilot: status

FixedHigh
Affected
  • Microsoft 365 Copilot for iOS before 2.107.2 and for Android before 16.0.19815.10000
  • Outlook for iOS, Android, and Mac before 5.2605; Teams for iOS before 8.3.1
  • Copilot-enabled Word, Excel, PowerPoint, OneNote, Loop, Power BI, and Edge mobile apps (fixed builds in Microsoft's update guide)
Status as of
Discovered by
Andi Ahmeti (Permiso Security)

What changed with Microsoft 365 Copilot

Microsoft's FAQ says an attacker could send an email crafted so that Copilot, when processing it, presents official-looking messages that push the user to click links leading to data exfiltration or a malicious site. Microsoft classifies the issue as information disclosure that requires user interaction, credits Andi Ahmeti of Permiso Security, and reported no public disclosure or exploitation at release.

The update guide lists fixed builds across Copilot-enabled apps, including Microsoft 365 Copilot for iOS 2.107.2 and for Android 16.0.19815.10000, Outlook for iOS, Android, and Mac 5.2605, Teams for iOS 8.3.1, plus Word, Excel, PowerPoint, OneNote, Loop, Power BI, and Edge mobile apps, and marks customer action as required.

How to fix it: Microsoft 365 Copilot

  1. 1. Update

    Update Copilot-enabled Microsoft 365, Outlook, Teams, Office, and Edge apps to the fixed builds listed in Microsoft's update guide.

    msrc.microsoft.com (opens in a new tab)

Advisories

Why Microsoft 365 Copilot matters for businesses

Indirect prompt injection turns an assistant's credibility into a phishing channel, because people tend to trust a Copilot summary more than the raw email. Keep Copilot-enabled apps current through device management, and teach staff that security alerts or links inside AI summaries deserve the same skepticism as the message itself.

Sources for this story

  1. Official
  2. CVE-2026-26133 (opens in a new tab)

    NIST National Vulnerability Database · nvd.nist.gov

    Official

Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.

  • #security
  • #cve
  • #microsoft-365-copilot
  • #prompt-injection
  • #phishing
  • #microsoft
  • Generated cover for Azure MCP Server by Microsoft: a shield and padlock over a hexagon lattice, in Microsoft brand colors.

    MicrosoftSecurity

    Microsoft patches Azure MCP Server SSRF that could leak its managed identity token

    In its March 2026 security release, Microsoft fixed CVE-2026-26118, an SSRF in Azure MCP Server Tools rated Important (CVSS 8.8). Someone who can talk to an MCP-backed agent could swap a resource ID for a malicious URL and capture the server's managed identity token. Fixed in 1.0.2 and 2.0.0-beta.17.

    Fixed

Want this working in your business?

We turn new AI capabilities into agents, voice receptionists, and automations you can actually use.