DEV Community

Serguey Shinder
Serguey Shinder

Posted on

Our Credit Controller Clicked Accept on a Real Microsoft Page and Resetting Her Password Changed Nothing

In March, one of our credit controllers received an email saying a customer had shared a remittance advice with her. The link led to a genuine Microsoft sign in page, and after she signed in, a genuine Microsoft screen asked her to let an app called Remittance Viewer read her mail and keep access to it. Customers do send remittances through odd services, so she accepted.

Nothing else happened that day. Her password was not stolen and nobody signed in as her. The app simply used the permission she had given it to read her mailbox through Microsoft's own interfaces, every day, for five weeks.

We found out when a builder rang to check an email asking him to pay his overdue balance into a new bank account. It came from a lookalike domain, quoted the right invoice numbers and repeated a conversation he had had with her two days earlier.

Our first response was the usual one. We reset her password, signed her out everywhere and checked her second factor. The mailbox kept leaking, because the app's access never depended on her password. She had granted it, and a grant to an app stays in place until someone removes it. We found it listed against her account in our directory, revoked it and deleted the app, and the leaking stopped.

Our tenant was set up in 2016 and still allowed any user to give any app access to their own data, as was normal then. Users can now consent only to apps from verified publishers asking for low risk permissions. Anything else becomes a request my team reviews, usually the same day. We went through every existing grant across the company and found two hundred and twelve. Most were harmless calendar and note tools, and we removed forty that read mail or files and that nobody could justify. Our incident checklist for a compromised mailbox now starts with app grants and mailbox rules, before the password.

Phishing no longer needs a fake page when the real one will ask for exactly what the attacker wants. Train people to read what they are permitting, and check what has been permitted already.

– Serguey Shinder

Top comments (0)