On Monday morning your office manager's inbox fills up with hundreds of newsletter sign-ups. Ten minutes later a Teams chat pops up from "IT Helpdesk": they've noticed the spam problem and can fix it if she opens Quick Assist and reads out a six-digit code. She's relieved someone is on it. She reads out the code.
That person isn't your IT provider. It's an attacker using a Microsoft 365 tenant of their own, and that code has just given them control of her PC.
This isn't a rare trick. Microsoft's threat researchers have published three separate investigations into it this year alone, in April, September and October 2026. The good news: it relies on default settings and good manners, not on a software flaw, so a few changes and one clear rule shut most of it down.
How the scam works
The pattern has been stable since Microsoft first described it in 2024, when a criminal group flooded inboxes with subscription emails, then phoned or messaged the victim on Teams posing as help desk staff. That campaign ended in Black Basta ransomware.
The steps today look much the same:
- First contact from outside. The attacker sets up their own Microsoft 365 tenant, gives it a name like "IT Support", and starts a Teams chat or call with your employee. Microsoft calls this cross-tenant helpdesk impersonation, with lures such as "Microsoft Security Update", "Spam Filter Update" or "Account Verification".
- A remote session. The employee is talked into opening Quick Assist, the remote help tool built into Windows, typing in a short code and approving the prompts. Microsoft notes that this step is often completed in under a minute.
- Quiet installation. While the user watches what looks like a fix, the attacker downloads an installer with an update-style name such as "devfix" or "Hotfix", which gives them a way back in after the session closes.
- Spreading and stealing. From that one PC they move to servers and domain controllers, install their own remote management software and copy business documents to external cloud storage.
Microsoft is explicit that in the October cases the activity did not exploit a vulnerability in Microsoft Teams. Everything the attacker used is a normal feature, used as designed. That's why your antivirus doesn't stop it at the front door.
Why it works on smart people
Being helpful to IT is what we teach staff to do. Three things make this lure especially convincing:
- It arrives inside a trusted tool. People have learned to distrust strange emails. A Teams chat feels internal, even when it carries an "External" label.
- The problem is real. If the attack starts with an inbox flood, the "helpdesk" is offering to fix something the employee can see happening.
- The tool is real. Quick Assist is made by Microsoft and is already on the PC. Nothing needs to be downloaded, so nothing looks suspicious.
Tighten Teams external access
By default, Microsoft documents that Teams is set to allow all external domains, so anyone with a Microsoft 365 tenant can find your staff and start a chat. For most small businesses, that's far more openness than they need.
What to do:
- Switch to an allow list. In the Teams admin center, under External access, allow only the domains of the customers, suppliers and partners you actually chat with. Microsoft's own guidance for these attacks is to restrict external access to trusted domains only.
- Decide on personal Teams accounts. Chat with unmanaged Teams accounts, meaning people using personal Teams rather than a company tenant, is a separate setting. Turn it off unless you have a clear business reason.
- Keep the warnings visible. When an external person contacts someone for the first time, Teams shows an accept or block screen. Tell staff that this screen is a decision point, not a formality. Microsoft has also announced brand impersonation warnings for first-time external callers, reported to be on by default. It's a useful extra, not a substitute for the allow list.
Take Quick Assist off the table
If your IT provider uses a different remote support tool, there's no reason for Quick Assist to be on your PCs. Microsoft's advice is to block or uninstall Quick Assist and other remote management tools you don't use, and if you use Intune Remote Help instead, to remove Quick Assist as a best practice.
What to do:
- Ask your IT provider which tool they use for remote support, and get it in writing.
- Uninstall Quick Assist on company PCs if it isn't that tool. Uninstalling is cleaner than blocking its network address, because Microsoft warns that blocking it also breaks Remote Help.
- Block other remote access apps your provider doesn't use, such as AnyDesk or TeamViewer, so the attacker can't simply switch tools.
Agree how real IT support contacts you
Settings narrow the door. A simple, shared rule closes it.
Your IT provider will never contact you out of the blue on Teams and ask you to open a remote session. If someone does, stop and call IT back on the number you already know.
Microsoft's guidance goes a step further and suggests an authentication phrase between the helpdesk and employees, so a genuine support person can prove who they are. For a small company, a short list works well: who your IT provider is, how they contact you, and what they will never ask for.
What to do:
- Write it down on one page, with your IT provider's real name, phone number and ticket address.
- Show staff what an external chat looks like, including the "External" label and the accept or block screen.
- Make reporting easy. A sudden inbox flood or an unexpected "helpdesk" chat should go straight to your IT contact. Suspicious messages can also be forwarded to suspicious@safeonweb.be.
If someone already let them in
Treat it as an incident, not an awkward mistake. Disconnect the PC from the network (pull the cable or turn off Wi-Fi) but don't switch it off, and call your IT provider. Microsoft advises to assume the attacker reached the network through that machine and rotate the credentials it could access, including administrator accounts. Our guide to the first hour of a ransomware attack walks through the next steps.
Where to start
- Check your Teams external access setting this week. If it allows all domains, build an allow list of the organisations you really chat with.
- Turn off chat with unmanaged Teams accounts unless someone needs it.
- Remove Quick Assist and any remote access tool your IT provider doesn't use.
- Send staff the one rule: IT never starts a remote session out of the blue, and an inbox flood followed by a "helpdesk" chat is an attack.
- Agree what happens if someone clicked: who to call, and that the PC goes offline straight away.
The takeaway
The fake helpdesk scam doesn't break into Microsoft 365. It's invited in through an open Teams door and a remote tool nobody needed. Narrow who can message your staff, remove the tools attackers ask for, and give everyone one clear rule about how real IT support behaves. That's an afternoon of settings and a five-minute team briefing, and it closes a route into company networks that Microsoft has had to warn about three times this year.