Active OAuth consent phishing drives cloud intrusions midweek as attackers bypass MFA
Marcus Webb ·
Security teams are fielding a midweek surge of phishing and business email compromise tied to malicious OAuth apps and session hijacking across Microsoft 365 and Google Workspace. The active campaigns, seen across North America and Europe, bypass multifactor authentication by abusing consent grants and long lived refresh tokens. Attackers aim to read mailboxes, plant inbox rules, and exfiltrate data at scale.
The technique matters because it is not a software flaw, it is an identity and configuration problem that survives password resets. Once a user or admin grants a rogue app Mail.Read, offline_access, or full_control scopes, the attacker gains durable API access until tokens are revoked and the app is removed.
Analysts continue to see adversaries register or compromise Azure AD applications, then request high value Graph API permissions through deceptive consent prompts. AiTM phishing kits such as EvilProxy and Greatness intercept cookies at login, which lets attackers replay sessions or bootstrap app consent from a hijacked context. Infrastructure is often hidden behind residential proxies and cloud front ends to blend with normal traffic.
Midnight Blizzard, tracked as APT29, has previously used OAuth applications and service principals to retain persistence in Microsoft 365 according to Microsoft disclosures. Criminal groups engaged in BEC also favor this approach because Graph API access enables silent mail forwarding and rule manipulation without noisy interactive sign ins. Google customers face the same pattern when attackers trick users into approving unvetted third party apps with broad Gmail and Drive scopes.
This is an active technique, not a new CVE. Security teams should treat it as a control plane intrusion that requires hardening the consent surface and revoking token based access, not only rotating passwords.
Mitigation path
- Disable or tightly restrict user consent to applications. Require admin consent for high risk scopes and use the Admin Consent Workflow.
- Restrict app registrations to vetted processes. Audit Enterprise Applications for unused or over privileged service principals and remove them.
- Enforce Conditional Access with authentication context and device requirements for Graph data access. Enable Continuous Access Evaluation and token protection where supported to reduce token replay.
- Monitor Entra ID AuditLogs and SignInLogs for new service principals, credential additions to app objects, and anomalous consent events. Alert on offline_access and Mail.Read across many users.
- Revoke refresh tokens and force reauthentication after any suspected consent abuse. For Google Workspace, enforce app access controls and restricted scopes, and review domain wide delegation.
For defenders, the broader implication is clear. Identity governance now sits on the front line, so cloud security programs must weight consent policy, token hygiene, and app governance at the same priority as patching and network segmentation.