Multi-Factor Authentication Alone Won’t Cut It

I frequently encounter a common misconception: “Multifactor authentication (MFA) protects against phishing”. In fact, only some forms of MFA are what is called phishing-resistant. Applications that periodically generate codes and push notifications do not protect against phishing. These second factors protect users if their password becomes known to attackers, often because they reused their password across multiple sites and one of them was the victim of a breach. They also protect users who have chosen a weak password that an attacker can brute-force.
However, these forms of MFA do not protect users when they are tricked into entering credentials into a phishing site. A phishing site will simply ask for the authenticator code in addition to the password, and then replay both the password and the code to the targeted site. The targeted site, being satisfied with the credentials presented to it, then lets the attacker in.
Rolling out Single Sign-On (SSO) and MFA enterprise-wide is a great first step in the journey to protect your users, but what comes next? Phishing-resistant MFA, as foreshadowed above, and protecting user sessions post-authentication where they live: in their browser.
Phishing-Resistant MFA
The Cybersecurity and Infrastructure Security Agency (CISA) refers to phishing-resistant MFA as the gold standard for MFA. Passkeys and physical security keys such as YubiKeys are common forms of phishing-resistant MFA. In a browser, these can be leveraged using the WebAuthn standard. The key innovation here is that these factors hold secrets that are cryptographically bound to a site’s origin and will only allow a user to authenticate to the domain for which they were configured.
When the user is tricked into attempting to authenticate to a look-alike domain, they cannot complete the authentication, making this form of adversary-in-the-middle attack impossible. This is because, despite their somewhat misleading name, adversary-in-the-middle attacks are unlike traditional man-in-the-middle attacks: the victim is not interacting with the target. They are interacting with the malicious site and the malicious site is replaying their interaction with the target on the side. Even though the victim failed to detect the difference, the phishing-resistant MFA can: the malicious site is unable to prove that it is the site for which the secret was created.
Beyond Phishing-Resistant MFA
In its Q2 2026 incident response report, Cisco Talos found that authentication abuse showed up in 65 percent of engagements, up from 35 percent the quarter before. Attackers used a range of techniques, including adversary-in-the-middle proxies, MFA fatigue attacks, stealing session tokens, and targeting OAuth device authorization grants.
As discussed, phishing-resistant factors will defeat adversary-in-the-middle attacks. They are also not vulnerable to MFA fatigue attacks. Attackers must now take it a step further by targeting the authenticated session, which is the second layer after MFA.
Once a user authenticates, the application issues a session token. This token is stored by the user’s browser, often in the form of a cookie. This session token is effectively a receipt saying authentication happened and is not required again until the session expires.
However, this implies that if the user’s browser is compromised, malware can steal this session token, and the attacker can then load it into their own browser and become authenticated without needing to provide a password and second factor. This makes protecting users from browser attacks paramount.
In an OAuth device authorization grant attack, an attacker tricks the victim into creating a secondary session, based on the session they have already in their browser. This type of interaction is commonly encountered when authenticating into streaming services on a new TV based on an existing session with the streaming service. Many business applications support a similar flow. The attacker now tricks the victim into providing a code generated within the existing session, no MFA required.
It is worth noting that at this point, resetting the victim’s password may not get the attacker out. First, the stolen session stays valid until it is revoked or expires, and resetting credentials does not necessarily invalidate existing sessions. Second, the attacker did not get in because they knew the password. They got in because of browser security weaknesses that remain unaddressed.
Closing the Session Gap: Takeaways For Security Leaders
The techniques described above have one thing in common: they play out inside the browser. The browser is where the session tokens live and where OAuth attacks take place.
Yet, most organizations’ security stacks watch the browser from the outside. Identity providers see the authentication, network tools see encrypted traffic, and endpoint tools see processes. Very few tools have visibility into the activity of pages loaded by the browser, creating a gap for bad actors to exploit.
Organizations should not stop at rolling out MFA. Beyond defeating authentication itself, attackers are increasingly targeting what happens after signing in. In practice, four steps can make a big difference:
- Adopt phishing-resistant factors: Passkeys and FIDO2 security keys will protect you from credential phishing attacks.
- Do not assume a stolen session is due to stolen credentials: Response playbooks should revoke active sessions and tokens for the affected account, then investigate how the attacker got in.
- Protect the browser: If an attacker can compromise the user’s browser, they can steal all their session tokens.
- Extend inspection inside the browser: Gain visibility into page-level activity.
MFA remains a foundational control at the authentication layer, but its scope ends where the session begins. If security stops at the login screen, attackers will keep walking through.
