Security

EvilTokens Device Code Phishing: Microsoft Guidance

3 min read

Summary

Microsoft has detailed how the EvilTokens phishing-as-a-service platform industrialized device code phishing to steal tokens and compromise more than 12,000 inboxes across 10,000 organizations. The report matters for security teams because it includes mitigation guidance, Defender XDR detections, and hunting recommendations to reduce risk from token theft and post-compromise persistence.

Need help with Security?Talk to an Expert

Introduction

Microsoft has published a deep dive into EvilTokens, a phishing-as-a-service platform that rapidly became a major threat in 2026. The campaign is notable because it abuses the legitimate device code authentication flow to capture tokens, bypass traditional MFA expectations, and maintain access to Microsoft 365 accounts.

For IT and security administrators, this is an important reminder that modern phishing increasingly targets session tokens and OAuth flows, not just usernames and passwords.

What’s new in Microsoft’s findings

Microsoft says EvilTokens:

  • Compromised more than 12,000 inboxes across over 10,000 organizations worldwide
  • Used AI-assisted phishing lures and inbox analysis to identify high-value targets
  • Abused the device code flow to trick users into authorizing attacker-controlled sessions
  • Enabled post-compromise activity such as email exfiltration, malicious inbox rules, and Microsoft Graph reconnaissance
  • Supported large-scale campaigns with short-lived infrastructure and automated backend logic designed to evade detection

Microsoft also linked the development and support of the platform to Storm-2992 and said its Digital Crimes Unit helped disrupt infrastructure used by the service.

Why device code phishing is effective

Device code authentication is designed for devices like smart displays, Teams devices, and conference systems that cannot complete a normal interactive sign-in. Attackers exploit this by initiating the sign-in themselves, then convincing a user to enter the code through a phishing lure.

Because the user completes a legitimate Microsoft authentication step, the attacker can obtain access without stealing the password directly. That makes this technique especially dangerous in environments that rely heavily on MFA as a primary defense.

Impact on administrators

Organizations using Microsoft 365, Entra ID, and Defender should review whether device code flow is truly required in their environment. Industries affected by EvilTokens campaigns included financial services, healthcare, higher education, construction, and real estate, showing that the threat is broad rather than niche.

Security teams should also assume that a successful token theft may lead to:

  • Persistent mailbox access
  • Hidden inbox rules
  • Internal reconnaissance through Microsoft Graph
  • Follow-on business email compromise activity

Microsoft’s guidance is clear:

  • Block device code flow wherever possible
  • If required for Teams devices, scope exceptions narrowly to specific resource accounts
  • Review Conditional Access configuration, including exclusions for the Device Registration Service only where needed
  • Use Microsoft Defender XDR detections and hunting queries from the Microsoft report
  • Strengthen mail flow rules, spoof protection, and connector configuration to reduce phishing delivery
  • Train users to be cautious of requests to enter device codes outside expected sign-in scenarios

Bottom line

EvilTokens shows how attackers are operationalizing token theft at scale using AI, automation, and legitimate authentication workflows. For defenders, the priority is to reduce unnecessary device code usage, tighten Conditional Access, and improve detection for suspicious token-based activity.

Need help with Security?

Our experts can help you implement and optimize your Microsoft solutions.

Talk to an Expert

Stay updated on Microsoft technologies

Securityphishingdevice code phishingMicrosoft Defender XDREntra ID

Related Posts

Security

Microsoft Digital Defense Report 2026: Key Security Insights

Microsoft's 2026 Digital Defense Report highlights how AI and growing system interconnectedness are reshaping both cyberattacks and defense strategies. The report emphasizes that organizations must secure AI, identities, data, and cloud environments together while improving signal correlation across tools to detect modern threats faster.

Security

Government Cyber Risk in 2026: Microsoft’s 5 Priorities

Microsoft says government agencies were the most targeted sector in 2026, accounting for 27% of observed cyber threat activity. The company urges public-sector leaders to focus on five resilience priorities, including faster response, AI security, bidirectional information sharing, and planning for incidents that spread across suppliers and essential services.

Security

Microsoft Ignite 2026 Security Guide: Key Sessions

Microsoft has published its security guide for Microsoft Ignite 2026, highlighting AI-first security themes, a dedicated Security Pre-Day, and technical sessions focused on securing identities, data, devices, clouds, and AI agents. For IT and security teams, the event offers an early look at Microsoft’s roadmap and practical guidance for building an AI-ready security strategy.

Security

CVE-2026-73570: Zimbra Mail Server Exploitation

Microsoft is tracking active exploitation of CVE-2026-73570, an unauthenticated command injection flaw affecting internet-facing Zimbra mail servers with the optional zimbra-snmp package installed and SNMP notifications enabled. The issue can lead to web shell deployment, privilege escalation, mailbox data theft, and persistent remote access, making immediate patching and configuration review critical for administrators.

Security

Phishing Abuses RMM Tools for Persistent Access

Microsoft security researchers observed phishing campaigns in July 2026 that used a legitimate MSP360 RMM installer disguised as meeting invites, PDF updates, and other lures to gain remote access. Attackers then deployed ConnectWise ScreenConnect for redundant persistence, highlighting the need for tighter controls on remote management tools and better detection of unapproved RMM activity.

Security

Azure DevOps Attack Path Exposed in New DART Report

Microsoft’s latest DART cyberattack report shows how a single compromised identity was used to access Azure DevOps, alter pipelines, and harvest Kubernetes credentials. The case highlights how tightly connected identity, DevOps, and cloud environments can let attackers move far beyond source code, making stronger identity and pipeline controls essential.