EvilTokens Uses Browser-Decrypted Phishing to Gain Microsoft 365 Account Access

· Original article ↗

Summary

EvilTokens hides phishing content with AES-GCM encryption and uses Microsoft’s device login flow to gain account access. The analysis traces its browser requests and shows how investigators can uncover related activity.

Key points

  • EvilTokens encrypts its landing-page HTML with AES-GCM, so static URL and network checks may miss the content shown after browser execution.
  • The phishing flow requests a user code and verification URI, then displays the code to persuade victims to authorize access through Microsoft’s legitimate device login flow.
  • The page polls the device-code session until authorization completes, then redirects the victim to OneDrive.
  • The analysis identifies gate-check, device-start, and session-status endpoints used to control and monitor the flow.
  • Investigators can use DOM changes, browser requests, signatures, and selected indicators to analyze the activity and pivot to related phishing campaigns.
  • The article reports recent activity concentrated in the United States and Europe, targeting organizations across several industries.

Article Details

Attack Vectors
  • EvilTokens uses a phishing page whose HTML is encrypted with AES-GCM and becomes visible after browser-side decryption.
  • The phishing flow prompts victims to complete Microsoft's legitimate device login flow, obtaining account access without directly stealing the victim's password.
  • The page requests a user code from the backend, displays it to the victim, and polls the device-code session status until authorization completes.
Defensive Notes
  • Inspect browser-rendered DOM changes and browser-level HTTP requests to reveal content and behavior hidden from static URL or network-level analysis.
  • Review the device-code flow endpoints and session behavior, and validate the associated domain, URI, and hash before using them in detections.
  • Avoid blocking or detecting shared CloudflareNet infrastructure solely on the basis of an observed IP address, as this could affect legitimate services.
  • Use decrypted page code, browser requests, endpoints, and behavioral patterns to develop phishing signatures, threat-hunting hypotheses, and detection rules.

Indicators of compromise

TypeIndicatorContext
DOMAINemp01825[.]workers[.]devDomain identified by the article as a specific EvilTokens session indicator for validation and detection.
MD5fcd1b654a0b3e8f85ca7cfdafe494d4bMD5 hash identified by the article as an indicator from the EvilTokens session for validation and detection.

MITRE ATT&CK

Vendors

Products

Tools

Countries

Industries

Related Articles