Sit down for a minute, because this one trips up even people who have worked in IT for years. You probably enabled multi-factor authentication (MFA) on your office account and thought, fine, that is the account sorted. Mostly true. But “MFA is on” is a statement about whether a second step exists, not about how hard that second step is to fool. Those are very different things.
What a “factor” really is
Authentication factors come in three families: something you know (password, PIN), something you have (phone, hardware key, smart card) and something you are (fingerprint, face). MFA means combining at least two different families. Two passwords is not MFA, because both are things you know.
Now, inside “something you have” there is a huge range of strength. An SMS code, a six-digit code from an authenticator app, a “Tap Approve” notification, and a physical security key all count as the same factor family. Their resistance to attack is nowhere near equal. Keep that thought; the rest of the article hangs on it.
How push-bombing works
Suppose an attacker already has your password. Perhaps it leaked in an old breach and you reused it, or you typed it into a fake login page. On its own the password is now useless to them, because your account asks for a second step.
So they simply try to log in. Your phone buzzes with a prompt: “Approve sign-in?” You did not do that, so you tap Deny. They try again. And again. Twenty prompts in ten minutes, at 11 p.m., while you are trying to sleep. This is called MFA fatigue or push-bombing. The attacker is not breaking the cryptography; they are betting that you will eventually tap Approve just to make the buzzing stop, or by accident, or because you assume it is some glitch in the system.
A variant adds a human touch. The attacker phones or messages you, pretending to be the IT helpdesk, and says “You will get a prompt, please approve it so we can fix the problem.” Now the approval feels legitimate. It is a social engineering attack wrapped around a technical one.
Compare this with credential stuffing, where the attacker replays leaked username and password pairs against many sites. Stuffing needs no cooperation from you at all, and MFA usually blocks it cold. Push-bombing needs you to cooperate, and that is exactly the gap it exploits.
What it looks like in logs
The pattern is quite distinctive if you know to look for it. In the identity provider’s sign-in logs, you see a successful password check followed by a burst of failed or denied second-factor attempts for one user, often from an IP address or country the user never uses. Then, sometimes, one approved attempt. That last line is the one that matters: a long run of denials followed by a sudden success, from a location that does not fit, is a strong sign of fatigue succeeding.
Credential stuffing looks different: many users, each with one or two failures, usually from a small set of source addresses, spread over a long window. One user and many prompts versus many users and few attempts. That difference in shape is how analysts tell the two apart.
A ladder of second factors
Think of second factors as rungs on a ladder, weakest at the bottom.
- SMS codes. Better than nothing. But SIM-swap fraud, where a criminal gets a duplicate SIM issued in your name, and message interception both exist, and a fake login page can simply ask you for the code and relay it.
- Plain approve/deny pushes. Convenient, but exactly what push-bombing targets.
- Number matching. The login screen shows a two-digit number and your phone makes you type it. A blind tap is no longer possible, which defeats most fatigue attempts. A convincing fake login page can still trick someone, though.
- One-time codes from an authenticator app. No prompt to spam, but a user can still be talked into typing the code into a fake page, and the attacker uses it within its short window.
- Phishing-resistant methods: FIDO2/WebAuthn security keys and passkeys. The key signs a challenge that is tied to the real website’s address. A look-alike domain gets a different address, so the key simply produces nothing useful for it. There is no code for you to be fooled into typing.
That last point is the real shift. Every method below it relies on you noticing something is wrong. The top rung does not rely on your attention at all.
What organisations can do
Turn on number matching or move to an authenticator that shows context (application name, location). Put a limit on repeated prompts so that after a handful of denials the account is locked or the user is forced onto another method. Alert the security team on “many denials then an approval.” Teach people one rule that fits on a sticky note: if you did not just try to log in, never approve, and report it, because the denials themselves are the evidence that your password is out there.
For administrators, finance staff and anyone who can move money or change access, plan for hardware keys or passkeys. For ordinary staff, a good authenticator with number matching is a reasonable middle. Be honest with leadership that “we have MFA” is a starting line, not a finish.
One more thing to check on your own accounts today: whether your recovery route, such as a backup phone number or an email fallback, is weaker than your main login. Attackers love to ask for the easy door.
Leave a Reply