ProBackend
Threats & Attacks
5 days ago5 min read

MFA bypass attacks: What EvilProxy’s Microsoft 365 campaign teaches

MFA bypass attacks: What EvilProxy’s Microsoft 365 campaign teaches

Can MFA be bypassed? EvilProxy’s reverse-proxy phishing demonstrates how stolen session cookies can defeat MFA—and what organizations can do to reduce account-takeover risk.

Can MFA be bypassed? EvilProxy shows how

Yes—some attacks can get around the protection MFA is meant to provide, without guessing or defeating the second factor itself. EvilProxy’s reverse-proxy phishing illustrates the distinction: a victim completes a real authentication flow, including the MFA challenge, while an attacker’s server relays the exchange and captures the resulting authenticated session cookie. The attacker can then reuse that session. MFA worked as designed at login; the session became the prize. BleepingComputer and Proofpoint describe the campaign and its mechanics.

That’s why “MFA enabled” is not a complete account-security story. MFA remains valuable: it raises the bar against stolen-password attacks. But authentication can be abused through social engineering, and a session that has already been authenticated may be usable without prompting for the factor again. EvilProxy is a concrete example of an MFA bypass technique, not proof that every MFA method is equally vulnerable or that MFA should be abandoned.

What happened in the campaign

BleepingComputer reported researchers observed 120,000 phishing emails sent to users at more than a hundred organizations. The campaign targeted Microsoft 365 accounts and used EvilProxy, a phishing-as-a-service platform designed to target MFA-protected accounts. That scale matters. This wasn’t simply one convincing email sent to one executive; the service enabled a broad campaign while tailoring parts of the experience to targets.

Proofpoint reported observing the campaign from March 2023, with emails impersonating familiar brands including Adobe, DocuSign, and Concur. A message dressed as routine business activity has a built-in advantage: a recipient may expect a document, signature request, or expense workflow. Brand familiarity can make a link feel ordinary. It does not make the destination safe.

The campaign also used layered redirection. A click could pass through open redirects associated with YouTube or SlickDeals and then through additional redirects before reaching the phishing page. Those detours complicate inspection and can make the final destination less obvious to a user or automated scanner. The final page was designed to resemble the target organization’s Microsoft 365 login experience.

How EvilProxy captures an authenticated session

EvilProxy works as a reverse proxy between a person and the real service. Instead of merely showing a fake sign-in form and collecting a password, the phishing infrastructure relays requests to the legitimate login service. The user sees a familiar authentication process and supplies credentials; if prompted, they also complete MFA. The proxy sits in the middle of that interaction.

After successful authentication, the service issues a session cookie to the browser. Proofpoint explains that EvilProxy can capture authentication cookies as the login exchange passes through its proxy. A stolen cookie can let an attacker act within the authenticated session, bypassing the need to repeat the MFA challenge for that session. This is the core of the “MFA bypass” description: the criminal isn’t necessarily cracking the one-time code or push approval. The criminal is stealing the proof that the user has already authenticated.

The phishing flow was made more targeted, too. Proofpoint described attackers using a target’s email address to direct the person to a phishing page tailored to their organization. That extra personalization can lower suspicion: a page that appears to match the user’s employer feels more plausible than a generic login screen. Yet the user’s email and employer branding are not proof of legitimacy. The route taken to the page matters.

Why high-value accounts were a priority

Proofpoint reported that attackers were selective about which compromised accounts they pursued through account takeover, prioritizing VIP targets and disregarding some lower-ranking cases. Among breached targets, its breakdown included 39% C-level executives, 9% CEOs and vice presidents, and 17% chief financial officers; the remaining compromised users included employees with access to financial assets or sensitive information.

Those figures are a reminder that account compromise is not just an IT inconvenience. A senior leader’s mailbox or a finance employee’s account may expose sensitive conversations, business information, or financial workflows. The sources describe targeting and account takeover, not a confirmed loss amount or a specific downstream breach for every victim. Still, the risk is plain: an attacker who takes over a cloud identity may inherit access associated with that identity.

Proofpoint also reported that after compromising an account, the attackers added their own MFA method, described as an Authenticator App with Notification and Code, to establish persistence. That step changes the response calculus. A password reset alone may not address every foothold if an attacker has altered authentication settings. Incident responders should treat unexpected MFA registrations as a serious signal and examine the account’s authentication and session state.

What organizations can do about MFA bypass attacks

Start with phishing-resistant authentication. Proofpoint recommends FIDO-based physical security keys. These are a stronger choice for resisting credential-phishing scenarios than relying on users to recognize every carefully staged login page. The campaign shows why simply asking employees to “look closely” is an inadequate control: the victim can be interacting with a convincing, relayed sign-in experience and still expose a session.

Email controls matter as well. Proofpoint recommends stricter email filtering, and its campaign description gives defenders concrete patterns to consider: brand impersonation, suspicious links, redirect chains, and messages that steer users to an unexpected Microsoft 365 sign-in. Filtering should be paired with a clear reporting route so employees can flag messages without needing to prove the link is malicious themselves.

Organizations should also review identity controls around new authentication methods and session handling. The observed persistence step—adding an attacker-controlled MFA method—means teams should investigate newly registered factors after a suspected compromise. Revoke active sessions and assess account changes as part of containment; do not assume changing a password alone ends an attacker’s access. These are prudent response steps prompted by the reported cookie theft and MFA-method registration, rather than claims that one control eliminates every attack.

Finally, keep the message proportionate: MFA is still worth using. It stops many attacks that depend on passwords alone. The lesson from EvilProxy is to pair MFA with phishing-resistant options, email defenses, careful monitoring of authentication changes, and a fast process for responding to suspected account takeover. Security controls work best in layers, especially when the attacker’s trick is to make the real login process part of the phishing operation.

The practical answer

Can MFA be bypassed? In a session-theft attack such as EvilProxy, yes: attackers can relay a sign-in, let the user complete MFA, and steal the authenticated session cookie. That does not make MFA pointless. It means organizations should protect the authentication session as well as the password and factor, favor phishing-resistant authentication where available, and respond quickly when an account’s sessions or MFA methods change unexpectedly.

mfa be bypassed? evilproxy shows howmfa be bypassed? evilproxy shows how