Huntress Links Microsoft 365 Device-Code Phishing Campaign to EvilTokens

· Original article ↗

Summary

Huntress says an EvilTokens campaign used Railway-hosted infrastructure for Microsoft 365 device-code phishing, affecting 344 organizations. The investigation details token replay and phishing methods; Huntress reports blocking further attempts.

Key points

  • Huntress and Flare attributed the campaign to the EvilTokens phishing-as-a-service platform; Huntress says it first observed compromises on February 19, 2026.
  • The campaign targeted Microsoft 365 identities at 344 organizations across the US, Canada, Australia, New Zealand, and Germany, with device-code phishing identified in 87.4% of cases.
  • Attackers used Railway infrastructure to harvest and replay OAuth tokens. Huntress says three Railway IPs accounted for about 84% of observed authentication events.
  • Phishing delivery included personalized lures, multi-hop redirects, email security vendor URL rewriters, compromised websites, and Cloudflare Workers-hosted pages.
  • Huntress reports that EvilTokens offers phishing products with AI-assisted lure and email-filter evasion features; it did not directly observe attackers using Railway’s AI features.
  • Huntress blocked Railway-originated logins for eligible tenants and says it had blocked 113 attempted compromises by March 23, in addition to about 350 reported over the prior two weeks.
  • Recommended responses include investigating sign-ins from Railway IP ranges, revoking affected users’ refresh tokens, restricting device-code authentication, and enabling Continuous Access Evaluation.

Article Details

Attack Vectors
  • Device-code phishing tricked Microsoft 365 users into entering attacker-generated codes at the legitimate Microsoft device authorization endpoint, yielding OAuth access and refresh tokens.
  • Phishing lures used personalized construction bid, document-sharing, voicemail, and other pretexts, with links wrapped in legitimate email-security URL rewriters and multi-hop redirects.
  • Phishing pages and redirect hops were hosted on Cloudflare workers.dev, Railway, third-party platforms, and compromised organization websites.
  • The attackers replayed harvested tokens from Railway infrastructure; BAV2ROPC user-agent events were associated with automated refresh-token exchanges.
  • Huntress attributed the campaign to the EvilTokens Phishing as a Service platform, whose advertised features include phishing-lure customization and credential capture.
Defensive Notes
  • Hunt for successful sign-ins from the listed Railway IPs; the article says any successful authentication from this infrastructure is a confirmed compromise.
  • Revoke affected users’ refresh tokens. The article cautions that in-flight OAuth access tokens may remain valid for about an hour unless Continuous Access Evaluation is enabled.
  • Consider blocking Railway IP ranges with Conditional Access Named Locations where no legitimate Railway-deployed applications are used.
  • Restrict device-code authentication flows to identities that require them, and require compliant devices for Exchange Online and SharePoint.
  • Enable Continuous Access Evaluation to reduce token-revocation latency, and train users not to trust device-code prompts solely because they appear on the legitimate Microsoft authentication endpoint.
  • Monitor for Railway sign-ins using BAV2ROPC and for iPhone user-agent strings pairing iOS 18 with Version/26.x from cloud-provider IP ranges.

Indicators of compromise

TypeIndicatorContext
DOMAINapp[.]usewhale[.]ioThird-party document-creation site identified as used to host a phishing lure.
DOMAINr[.]brandreward[.]comLink-tracking and redirection host identified as used to hide phishing-link destinations.
DOMAINtrack[.]tec35[.]comLink-tracking and redirection host identified as used to hide phishing-link destinations.
DOMAINwww[.]hits2babi[.]comLink-tracking and redirection host identified as used to hide phishing-link destinations.
DOMAINwww[.]taskade[.]comThird-party site identified as used to host a phishing lure.
IPV4162[.]220[.]232[.]223Railway IP listed as a source of OAuth activity, active as of March 19 across one organization.
IPV4162[.]220[.]232[.]235Railway IP associated with SAML and OAuth token activity.
IPV4162[.]220[.]232[.]55Railway IP listed among the campaign’s OAuth indicators.
IPV4162[.]220[.]232[.]57Railway IP associated with token-replay activity.
IPV4162[.]220[.]232[.]99Railway IP identified as a source of SAML authentication replay.
IPV4162[.]220[.]234[.]161Railway IP listed among the campaign’s authentication indicators.
IPV4162[.]220[.]234[.]34Railway IP listed among the campaign’s authentication indicators.
IPV4162[.]220[.]234[.]41Railway IP identified as the primary token-attack engine and a source of attacker authentication activity.
IPV4162[.]220[.]234[.]66Railway IP associated with device-code and token-replay activity.

MITRE ATT&CK

People

Vendors

Products

Tools

Countries

Industries

Related Articles