An MFA fatigue attack floods a user with push authentication prompts until one gets approved out of confusion, annoyance, or habit. The attacker starts with a stolen password, and the second factor, meant to stop them, becomes the door they walk through. Microsoft made number matching the default in Authenticator in May 2023 because of it.
The Uber case, in one paragraph
In September 2022, a member of the Lapsus$ group got hold of an Uber contractor's credentials and started sending push authentication prompts. The contractor, worn down by a stream of notifications over several hours and assuming IT was behind it, approved one. That single tap gave the attacker access to Uber's internal network, and the breach that followed was one of the most publicized of the year. The attacker did not break the cryptography. They broke the person.
The attack works because the second factor is a prompt, and a prompt is a decision, and a decision made a hundred times in a row is a decision made badly. The user who gets one push notification reads it carefully. The user who gets fifty push notifications in an hour stops reading them and starts tapping Approve to make them stop. The attacker knows this, which is why the attack is not one prompt but a flood.
Related reading on this site: Home and Invalid Clicks and ROAS: When Bot Traffic Eats Your Ad Budget cover the neighbouring problems in more depth.
Why one tap is enough
One tap is enough because the attacker already has the password. The push prompt is the only thing standing between them and the account, and the prompt is a yes-or-no question. A yes opens the account. A no keeps it closed, but the attacker can send another prompt, and another, and another, until the user says yes out of confusion, annoyance, or habit. The attack does not need the user to be tricked. It needs the user to be tired.
The scale of the credential problem makes the attack cheap to run. Akamai counted 26 billion credential stuffing attempts per month in 2024, and Verizon found stolen credentials were the initial access in 22 percent of confirmed breaches in 2025. The attacker who has a password is not guessing. They are working through a list, and the push prompt is the last gate on every account on that list.
Number matching and the end of blind approval
Number matching is the fix, and it is simple: instead of a blind Approve button, the login screen shows a number, and the user has to type that number into the authenticator app. A user who is being fatigued will see a number they do not recognize, because the number belongs to the attacker's login attempt, not theirs. The tap that used to be mindless becomes a match, and a match that does not make sense is a match the user will not complete. Microsoft enforced number matching by default in May 2023.
The rest of the defense is rate limiting and monitoring. Lock the account after repeated denials, alert on a burst of prompts, and move privileged users to phishing-resistant MFA. The fatigue attack is a human problem with a technical fix, and the fix is to make the approval decision impossible to make by accident. A prompt that can be approved by tapping the same button a hundred times is a prompt that will be approved by accident, and the attacker is counting on it.
Frequently asked questions
What is MFA fatigue?
An attack where someone who already has your password triggers repeated multi-factor authentication prompts until you approve one out of confusion or annoyance. It is also called push bombing or MFA bombing, and MITRE tracks it as T1621.
How did the Uber breach happen?
In September 2022, a member of the Lapsus$ group sent an Uber contractor a stream of push notifications over several hours. The contractor, worn down and assuming IT was behind it, approved one. That single tap gave the attacker access to Uber's internal network.
Does MFA fatigue break the cryptography?
No. The cryptography holds. The attack works on the human, not the math. The second factor stays intact, and the attacker only needs the user to approve a single prompt to get in.
What stops MFA fatigue?
Number matching, which replaces blind Approve with a code the user has to type, is the main fix. Microsoft enforced it by default in May 2023. Rate limiting prompts, locking accounts after repeated denials, and moving privileged users to phishing-resistant MFA complete the defense.