Stolen Credentials for the Microsoft Outlook Add-in The first known example of a malicious Microsoft Outlook add-in being used against users in real-world situations has been found by security researchers. More than 4,000 Microsoft account credentials, credit card numbers, and responses to banking security questions were stolen from a compromised meeting scheduler called AgreeTo. I concur.In December 2022, it was released as a valid open-source project on the Microsoft Office Add-in Store.

Agree's Koidex reportTo include (Source: Koi AI) It was a useful meeting tool that received good feedback. But in the end, the developer gave up on the project and erased the related Vercel deployment. As a result, the hosting URL for the add-in (outlook-one.vercel.app) became orphaned and open for registration. An attacker set up a phishing kit and claimed the accessible URL.

The Office Add-ins marketplace's AgreeTo page (Source: Koi AI) Anyone who still had the add-in installed or had downloaded it from scratch saw the attacker's phishing page appear directly in the trusted Outlook sidebar because the add-in was never taken down from Microsoft's store. The technical error Dynamic Dependencies at a Distance The Office add-ins' architecture was exploited in the attack. Add-ins are "remote dynamic dependencies," which are essentially XML manifests that load a URL in an iframe, as opposed to conventional installed software.

Description of the Stage The Gap Microsoft only examines the manifest when it is first submitted (2022 in this instance). Microsoft does not regularly check that live content is hosted at the authorized URL.

The Outcome Without starting a fresh security review, the attacker replaced the scheduling tool with a phony Microsoft login page. Victims were asked to log in when they opened AgreeTo. Their IP addresses and login credentials were collected by a script, which then sent the information to the attacker through a Telegram bot.

Add-in Shows False Login (Source: Koi AI) After locating the attacker's inadequately secured exfiltration channel, Koi Security was able to uncover the campaign. The complete dataset of 4,000 victims, including banking information and security responses directed at Canadian institutions, was recovered. Technically, the attacker could read and alter user emails because the add-in had "ReadWriteItem" permissions. But credential harvesting was the attack's main objective.

In response to the report, Microsoft removed the add-in.

This incident brings to light a crucial supply chain risk: if the infrastructure supporting trusted software is abandoned, it can subtly turn malicious years after it has been approved. X, LinkedIn, and LinkedIn for daily ZeroOwl. To have your stories featured, get in touch with us.