TL;DR
- Consent phishing tricks a user into approving a malicious OAuth app, so the attacker never needs a password or MFA code.
- Access persists after a password reset, because the attacker holds a token and not your credentials.
- Spot it by looking at new, unverified apps and the permissions they request—not just at sign-ins.
Recently, Expel’s SOC has seen an increase in consent phishing incidents. Consent phishing is tricky because victims don’t type passwords into a fake page—they sign in on a real account page, and that’s the whole attack. If credential phishing is stealing the keys, consent phishing is convincing you to cut a new one and then hand it over.
What is consent phishing?
Consent phishing attacks operate by leveraging legitimate authorization infrastructure to gain persistent access to resources. In contrast to credential compromise, these attacks side-step the login process altogether, targeting accounts with direct access to personal or corporate data. Unsuspecting users consent to the access/permissions requested because the apps are platformed on legitimate services (i.e. Microsoft and GitHub).
Consent phishing works because the page is real. The consent screen is hosted by the legitimate provider, and the bad actor gets token access, no password needed.
How does consent phishing work?
- The initial attack vector is typically a spearphishing email.
- The phishing link directs the user to a legitimate consent page, like Microsoft or Google Workspace.
- If the request is accepted, the actor receives an authorization code from the application and can generate a new session token, gaining access to the victim’s environment.
- The new token then allows API calls to interact depending on the permissions granted. Attackers can potentially gain access to sensitive data, manipulate API calls, exfiltrate data, and more.
The only way to revoke access once it’s gained through consent phishing is by finding and removing the malicious app—a password reset won’t work.
[IMAGE]
Two ways attackers use consent phishing
From what the cybersecurity industry has seen so far, there are two types of consent phishing we see most often:
- The classic approach: Attackers target a platform (like GitHub) by posting fake security alerts that point to a malicious app (in this case, an OAuth app called “gitsecurityapp”, which expands a user’s access. In 2025, this style attack touched 12,000 repositories. In this scenario, once access is approved, the attacker can modify repos, push malware, and take sensitive data because of the hidden app’s granted permissions.
- Consent phishing with a twist: In this type of attack, bad actors aren’t looking for just consent—they’re setting up an opportunity to steal session, credential, and MFA code details if the victim authenticates the request. In this scenario, attackers impersonate tools like Adobe or DocuSign they’ve infected with malicious Microsoft OAuth apps.
How does Expel detect consent phishing?
Our MDR service has detection rules for new/unauthorized app access, suspicious API calls, and anomalous session tokens.
On the email side, our phishing team has received multiple email reports leveraging consent phishing URLs.
Protecting against consent phishing
Microsoft has provided guidance on staying secure against consent phishing. They recommend:
- Locking down user consent. Configure settings so users can only approve apps from verified publishers or your own org, and only allow approval for low-risk permissions.
- Auditing what’s already approved. Routinely review apps and consented permissions, and enforce least privilege.
- Not trusting names or domains. Attackers can spoof both, so always check the publisher and read the permissions requested, every time.
- Reporting weird things. Microsoft allows users to report suspicious apps directly in consent prompts.
- Knowing what disabling does (and doesn’t) do. Even when Microsoft disables a malicious app, existing tokens are still valid until they expire, so don’t forget to revoke them.
Indicators of compromise (IOCs)
- New applications, unverified applications, applications requesting read/write access
- Unexpected emails with a call to action (“Update security now, “new app for compliance”)
- API traffic from commercial cloud providers instead of internal/typical infrastructure
