How attackers skipped MFA at 79 organizations by stealing sessions, not passwords

Between July 14 and September 22, 2026, hackers broke into Microsoft 365 accounts at 79 organizations without guessing a single password. Researchers at Huntress identified the campaign and published their findings on October 22, 2026, recording at least 497 successful sign-ins. The attackers didn't need passwords because they stole something more useful: live, already-authenticated sessions. That meant multi-factor authentication, the safeguard most IT teams lean on hardest, never came into play. If your phone or collaboration systems connect into Microsoft 365 or Teams, which many do, this one is worth understanding in full, not just skimming the headline.

The key points

  • Attackers stole already-authenticated sessions instead of passwords, which let them skip MFA entirely
  • The campaign ran 10 weeks, hit 79 organizations, and produced at least 497 successful sign-ins
  • Two phishing methods were combined: fake sign-in pages that relayed credentials, and device-code approval on Microsoft's real site
  • In some cases, attackers wiped and replaced MFA methods within three hours to lock out the real user
  • Conditional Access policies blocked over a thousand attempts in one wave, showing MFA alone isn't the full defense

The attack worked by stealing a session that had already cleared MFA

Most account-takeover defenses assume an attacker has to get past a password and then a second factor. This campaign sidestepped that assumption entirely. Stolen tokens let attackers reuse proof that a user had already completed multi-factor authentication, giving them access without ever repeating that challenge. In practical terms: the attacker wasn't breaking in. They were walking through a door the real user had already unlocked.

Huntress's investigation describes this as token theft and account abuse, not a named piece of malware. That distinction matters. There's no single bad file to detect and remove. The attacker's tool was a convincing fake login flow, not malicious code sitting on a device.


Two phishing methods did the actual session theft

The operation combined adversary-in-the-middle phishing, device-code phishing, and social engineering. These aren't the same thing, and it's worth knowing how each one worked.

In the relay-based attacks, a fake sign-in page sat between the victim and Microsoft. It forwarded the victim's real login and MFA challenge straight through to Microsoft, then captured the authenticated session that came back. The user thought they'd logged in normally. In some cases, attackers replayed that captured session from rented infrastructure run through a German hosting provider, 1337 Services GmbH.

The second method, device-code phishing, is more unsettling because the victim isn't on a fake page at all. Huntress observed this technique during the campaign's August wave: victims approved an attacker-controlled session while signing in on Microsoft's own legitimate website. There was no spoofed URL to catch, because there wasn't one.


Once in, attackers moved fast to lock out the real owner

Speed was a defining feature of this campaign. The gap between authentication and account takeover was sometimes just minutes: 12 minutes in one July incident, four minutes in the final takeover recorded on September 22. That's the most recent activity Huntress documented in this campaign, with no newer follow-up reported since.

Attackers also worked to keep the access they'd gained. They registered at least 48 rogue devices in Microsoft Entra and replaced victims' authentication methods outright. In 19 of the 79 organizations, attackers wiped existing MFA methods and registered replacements within three hours of their first sign-in. That's not just getting in. That's quietly taking over the account's own security settings before anyone notices anything is wrong.

Once inside, some attackers covered their tracks with inbox rules built to hide specific messages. In one case, a compromised account sent a payroll-related email to an external accountant, while rules were set up to hide banking-related messages and the replies to them. It's a reminder that the damage from this kind of access isn't only about data exposure. It can mean real financial fraud carried out from an account everyone trusts.


Defenses beyond MFA actually worked here

This isn't a story where nothing stopped the attackers. Conditional Access, which restricts sign-ins based on conditions like device, location, or risk level, blocked 1,052 attempts at four organizations during one campaign wave. That's a meaningful data point: a layer of policy enforcement sitting behind MFA caught what MFA alone didn't.

Huntress's own recommendations focus on the same idea: don't just trust that MFA was completed once. Revoke sessions, reset passwords, remove unauthorized devices and MFA methods, and review inbox rules for anything that looks like it's hiding messages. They also recommend restricting unnecessary device-code authentication and examining new application permissions, since both were part of how this campaign gained and kept access.

It's also worth being honest about what we don't know. Huntress hasn't named the attacker or the 79 affected organizations, hasn't said whether passwords were also captured alongside the tokens, and hasn't disclosed which countries the victims are in. I'm not going to fill in those blanks with guesses, and you should be skeptical of anyone who does.


What this means if your phone or collaboration tools plug into Microsoft 365

Here's where this stops being just a Microsoft story. A lot of the phone systems, meeting platforms, and collaboration tools mid-market companies run day to day are wired into Microsoft 365 or Teams for identity, calendar, presence, or single sign-on. That's normal, and usually a good thing for convenience. But it also means a compromised Microsoft 365 session can potentially reach further than the inbox. If a login integration trusts that session, it may not ask many questions of its own.

I'm not going to tell you your setup has this exposure, because I don't know your environment and TeleSwitch doesn't provide Microsoft 365 security directly. What I'd say, as someone who spends my time looking at how phone and collaboration systems connect into the rest of a client's stack, is that this is a genuinely good moment to ask the question. Not out of alarm. Just because the mechanism here, a stolen session standing in for a completed login, is exactly the kind of thing that can quietly cross from one system into another when they're connected. A closer look at how your sign-in and session policies apply across every connected system, not just Microsoft 365 itself, is time well spent.


TeleSwitch is a vendor-neutral technology advisor that helps businesses make smarter decisions around voice, cloud, connectivity, and security. We work with leading providers, but our loyalty is to our clients. With deep market knowledge and no-pressure guidance, we help you compare options, avoid costly mistakes, and choose solutions that support your long-term goals.

Next
Next

What Needs to Be in Place Before Your First AI Agent Goes Live