Skip to content

Does MFA stop phishing? What it blocks in Microsoft 365 and what it doesn't

Short answer

Mostly yes. A 2023 Microsoft study found MFA cut the risk of account compromise by 99.22% across the accounts it examined. But fake sign-in pages can relay a text or app code in real time, so authenticator apps beat SMS and FIDO2 security keys beat both.

Key takeaways

  • MFA stops the bulk of password theft. Turn it on everywhere before you worry about the edge cases.
  • Fake sign-in pages (adversary-in-the-middle phishing) can relay a code or approval, so not all MFA is equal.
  • Rank your options: FIDO2 key or passkey, then authenticator app, then SMS.
  • Block legacy sign-in protocols and roll MFA out in groups, starting with admins.

"We have MFA" is one of the most comforting sentences in IT, and it's mostly earned. It's also not the whole story. Here's what MFA actually stops, what slips past it, and what we'd roll out first if your 10, 50 or 200 staff are on Microsoft 365.

What MFA really stops

Start with the good news. It's big. In a 2023 research paper, Microsoft found that MFA reduces the risk of compromise by 99.22% across the entire population it studied, and by 98.56% for accounts whose credentials had already leaked. That's a study of accounts showing suspicious activity, not a promise about your office. It does explain why we start every identity conversation with "is MFA on?"

Reduction in risk of account compromise with MFA
  • All accounts studied99.22 %
  • Accounts with leaked credentials98.56 %

Source: Microsoft Research, "How effective is multifactor authentication at deterring cyberattacks?" (2023)

Microsoft's older number was higher. A 2019 post said MFA can block over 99.9 percent of account compromise attacks. We'd quote the newer figure, since it comes from measured data. Either way the lesson is the same: with a stolen password and no second factor, an attacker walks right in.

The attack that gets around it

Now the catch. In an adversary-in-the-middle (AiTM) attack, the link takes you to a fake Microsoft sign-in page. You type your password, the attacker passes it to the real service, and then shows you a forged MFA screen. You approve it, and the attacker captures the session. Microsoft's write-up of one real campaign says the attacker later replayed the stolen session cookie from a different IP address, with no password or MFA needed.

Nothing about this means MFA failed. MFA did its job here, and the person handed it over anyway.

So the fix is a kind of MFA the fake page can't relay.

  1. 01

    The link arrives

    A phishing email, often from an account the victim already trusts, with a link to a page that looks like a Microsoft sign-in.

  2. 02

    The fake page collects the password

    The attacker passes it straight to the real Microsoft service, which asks for a second factor.

  3. 03

    A forged MFA screen appears

    The victim sees what looks like the normal prompt and enters the code or approves the request.

  4. 04

    The attacker keeps the session

    Microsoft's write-up says the victim's MFA response let the attacker capture the session token. From then on the attacker doesn't need the password or MFA.

Notice that every step uses the person's own, correct behavior. They typed the right password and answered the right prompt. That's why more training alone isn't the answer, even though it helps.

Which kinds of MFA hold up

MFA methods compared for a Microsoft 365 office
MethodStops plain password theftSurvives a fake sign-in pageOur take
Text message (SMS) codeYesNo, the code can be relayedBetter than nothing. Move people off it when you can.
Authenticator app (code or approval)YesMostly no, the approval can still be relayedMicrosoft's study found apps outperform SMS. The sensible default.
FIDO2 security key or passkeyYesBuilt to resist this attackUse for admins and finance first.
Microsoft's research paper says dedicated MFA apps outperform SMS. Its 2026 incident write-up recommends FIDO2 security keys as phishing-resistant sign-in.

That last row is the one Microsoft points to. After describing a 2026 AiTM campaign, it recommends passwordless sign-in with FIDO2 security keys as the phishing-resistant option. A key is tied to the real site, so a lookalike page gets nothing useful out of it.

What sits around MFA

MFA is one layer. Microsoft's own advice after that incident goes further: use conditional access, especially risk-based policies, continuous access evaluation and anti-phishing controls, and keep hunting for suspicious sign-ins.

In plain terms that means the system also looks at where a sign-in comes from, how risky it looks and whether a session should still be trusted an hour later. A stolen session replayed from a new country should get challenged, not waved through.

It also means someone has to watch the logs. We'd rather you didn't depend on remembering to do that yourself.

What we'd roll out, in order

  1. 01

    Turn MFA on for everyone, admins first

    Microsoft's own advice is to roll it out gradually by role rather than flip it for the whole company on a Monday. Start with admins and anyone who handles money, then widen.

  2. 02

    Block legacy sign-in protocols

    Older protocols such as SMTP can't do MFA, so attackers go looking for them. Microsoft recommends blocking legacy authentication.

  3. 03

    Prefer the authenticator app over SMS

    Same effort for staff, and the research says the app does better.

  4. 04

    Put FIDO2 keys on admin accounts

    A handful of keys is a small cost for the accounts that can change everything else.

  5. 05

    Make changing MFA methods a re-challenge

    In the 2023 case the attacker quietly added their own phone as a method. Microsoft suggests requiring an MFA re-challenge for MFA updates.

  6. 06

    Plan for the day it gets through anyway

    Have the first-hour checklist ready: what to do after a Microsoft 365 account gets phished.

Mistakes we'd avoid when rolling MFA out

  • Switching it on for 100% of staff on a Monday. Microsoft's own advice is to roll out gradually by role. A help desk drowning in lockouts turns people against MFA.
  • Forgetting the service accounts. Shared mailboxes, scanners and old apps that use legacy protocols like SMTP can't do MFA. Find them first, then block legacy sign-in.
  • Having no way back in. Keep two emergency admin accounts with strong, stored credentials, excluded from the policy and checked every quarter.
  • Keeping SMS as the permanent answer. Use it to get everyone started, then move people to the Microsoft Authenticator app.
  • Calling it finished. MFA is one control. Someone still needs to look at sign-in logs and mailbox rules.

Most of this is dull project management, not clever security. Offices that run it as a two-week project with a named owner usually finish. Offices that treat it as a switch tend to turn it back off after the first lockout.

Where we fit

We set up MFA and single sign-on for Microsoft 365, Google Workspace and VPN access, and we've been doing managed IT in New York since 2001. Our detection side watches for what MFA can't (here's what MDR is and how it differs from antivirus): 24/7 threat detection and response for computers and Microsoft 365 email, powered by Huntress. If you want to see how the pieces fit, start with our MFA and single sign-on page or our identity threat detection page. If you're also sorting out how staff use AI tools, see how to write an AI policy your staff will follow.

Questions people ask

Is SMS MFA still worth using?

Yes, over nothing. Microsoft's research says authenticator apps outperform SMS, so treat SMS as a stepping stone and move people to the app, then to keys for admins.

What is an AiTM phishing attack?

Adversary-in-the-middle phishing puts a fake sign-in page between you and the real one. It relays your password and your MFA response to Microsoft and keeps the signed-in session for itself.

If MFA isn't enough, why bother?

Because the problem it solves is by far the most common one. Skip it and a leaked password is enough. Use it, then tighten the method over time.

Does approving a push notification stop an AiTM attack?

Not reliably. In the 2023 case Microsoft describes, the victim answered the forged MFA prompt and the attacker captured the session. Microsoft points to FIDO2 security keys as the phishing-resistant option. Push approvals are still better than SMS.

Sources

  1. Microsoft Research: How effective is multifactor authentication at deterring cyberattacks? (2023)
  2. Microsoft Security blog: One simple action you can take to prevent 99.9 percent of account attacks (2019)
  3. Microsoft Security blog: Detecting and mitigating a multi-stage AiTM phishing and BEC campaign (2023)
  4. Microsoft Security blog: Resurgence of a multi-stage AiTM phishing and BEC campaign abusing SharePoint (2026)
NS

Written by

Neteam SolutionsManaged IT and cybersecurity, NYC since 2001

Neteam Solutions Inc. runs IT departments and security for businesses in New York City, Nassau County and northeast New Jersey from offices in Astoria and Williston Park.

Last updated October 28, 2026