Microsoft Addresses Serious Vulnerability in Translator Hub
23.7.18 securityweek Vulnerebility

A serious vulnerability in the Microsoft Translator Hub could be exploited to delete any or all of the 13000+ projects hosted by the service, a security researcher has discovered.

The service allows interested parties to build their own machine translation system tailored for their organizational needs and then use it, via the Microsoft Translator Text API, in applications, websites, with Microsoft Document Translator, and more.

According to Microsoft, the Translator Hub allows enterprises build translation systems, and allows governments, universities and language preservation communities to “build translation systems between any pair of languages, including languages not yet supported by Microsoft Translator, and reduce communication barriers.”

While hunting for vulnerabilities on the Hub, security researcher Haider Mahmood discovered that the HTTP request for removing a project contained the “projectid” parameter, which is the ID of the individual project in the database.

Furthermore, Mahmood also discovered that the request also had no Cross-Site Request Forgery (CSRF) protection. This means that an attacker could exploit the CSRF vulnerability to impersonate a legitimate, logged in user and perform actions on their behalf.

An attack scenario, he says, would require for an attacker to know the ProjectID number of a logged in victim. Thus, they could include a URL in a page to issue a remove command and, as soon as the victim visits that page, the request would be sent from their browser and the project removed.

Further analysis of the issue revealed an Indirect Object Reference vulnerability, which could essentially allow an attacker to set any ProjectID in the HTTP project removal request and delete any of the projects in Microsoft Translator Hub.

In fact, by iterating through project IDs starting from 0 to 13000, an attacker could delete all projects from the database, the security researcher reveals.

Mahmood reported the vulnerability to Microsoft in late February 18. The company addressed the issue within the next two weeks, and also offered the researcher an acknowledgement on their Online Researcher Acknowledgement page.