Dev.to Security 🔐 Cybersecurity 👁 0 📖 4 min read

Verify Before You Reset, a Help Desk Callback Procedure for Password and MFA Requests

A help desk that resets a password or an MFA method on the strength of one phone call gives access to whoever sounds most convincing on that call. The fix is a callback procedure that runs before every reset, plus an SLA

A help desk that resets a password or an MFA method on the strength of one phone call gives access to whoever sounds most convincing on that call. The fix is a callback procedure that runs before every reset, plus an SLA that doesn't punish the technician for the minutes the check takes.

I run operations at a managed IT provider in Texas, and this is the procedure I'd ask any help desk to write down.

The problem

CISA published advisory AA23-320A on the Scattered Spider group in November 2023 and last revised it in July 2025. It describes the group placing calls to "convince IT help desk personnel to reset passwords and/or transfer MFA tokens."

The callers prepare first. By the time they phone the help desk, they already know the employee's name, job title and manager. A technician who checks the caller against those 3 facts will pass them every time, because those are the facts the attacker researched.

That's why identity questions the caller can answer don't count as verification here.

How the callback works

The procedure has 5 steps. None of them needs a new tool.

  1. The caller asks for a password reset, an MFA reset or a new MFA device.
  2. The technician doesn't reset anything on that call. They log the request and hang up.
  3. The technician calls the employee back on the number already in the file, not a number the caller gives.
  4. If that number doesn't answer, the employee's manager confirms the request through a second channel, such as Teams or email.
  5. Once the request is confirmed, the technician issues a Microsoft 365 Temporary Access Pass, which is a time-limited code the employee uses to set up a new sign-in method.

Nobody reads a password out over the phone at any point. The attacker who called in step 1 never receives anything, because the response goes to a number or a manager they don't control.

Why Temporary Access Pass instead of a new password

A new password read aloud has to travel through the conversation, which is the channel the attacker is in. A Temporary Access Pass is issued only after the callback succeeds, and the employee uses it to register a sign-in method of their own. It expires on its own, so a code that sits unused doesn't stay valid forever.

The point of the step is that the credential is created by the verified employee, not handed to whoever was on the first call.

Where it belongs in the SLA

A callback adds a few minutes to every reset ticket. If the SLA counts those minutes against the response clock, it quietly rewards technicians who skip the check. So the contract should say 2 things.

  • Every password or MFA reset requires a callback to a number already on file, or confirmation from the employee's manager through a second channel.
  • The time spent on that verification is excluded from the response clock.

A weak contract says nothing about resets at all, which leaves the technician to decide on the call.

When a reset request is really an incident

Some reset calls are the first sign of an incident. A user says the sign-in page looks different, or a folder of spreadsheets won't open. Once a business determines that a breach happened, legal clocks start.

  • Texas Business and Commerce Code Section 521.053 gives a business 60 days to notify affected people, and 30 days to notify the attorney general when 250 or more Texans are involved.
  • DFARS 252.204-7012 defines rapid reporting as within 72 hours of discovering a cyber incident, for companies that carry it in defense contracts.

NIST's SP 800-61 Revision 3, published in April 2025, treats incident response as part of everyday cybersecurity risk management. For the help desk, that turns into 4 lines in the contract. One says who can declare a security incident. One sets how fast the customer's named contact gets a phone call. One requires logs to be saved before any machine is wiped. The last one says who notifies the cyber insurance carrier.

Gotchas

  • Speed promises work against you here. A help desk proud of 2-minute resets is describing the exact behavior the advisory warns about.
  • The callback number has to come from the file. If the technician dials whatever number the caller offers, the check verifies nothing.
  • Verification time inside the clock. If it isn't excluded in writing, the pressure to hit the number will eventually win.
  • Pauses need a log. If verification time is handled with a waiting-on-customer pause, the monthly report should list every pause with its reason, or a 4-hour fix can show up as 40 minutes.

Takeaway

Write the callback into the SLA before you write in a speed target. The check costs a few minutes per reset, and it removes the one step a social engineering call depends on.

📰 Read the original article on Dev.to Security

Originally published by Dev.to Security. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.