Computer Vision Annotation Tool (CVAT) is an interactive video and image annotation tool for computer vision. Starting in version 2.2.0 and prior to version 2.14.3, if an attacker can trick a logged-in CVAT user into visiting a malicious URL, they can initiate a dataset export or a backup from a project, task or job that the victim user has permission to export into a cloud storage that the victim user has access to. The name of the resulting file can be chosen by the attacker. This implies that the attacker can overwrite arbitrary files in any cloud storage that the victim can access and, if the attacker has read access to the cloud storage used in the attack, they can obtain media files, annotations, settings and other information from any projects, tasks or jobs that the victim has permission to export. Version 2.14.3 contains a fix for the issue. No known workarounds are available.
Weakness
The web application does not, or can not, sufficiently verify whether a well-formed, valid, consistent request was intentionally provided by the user who submitted the request.
Potential Mitigations
- Use a vetted library or framework that does not allow this weakness to occur or provides constructs that make this weakness easier to avoid.
- For example, use anti-CSRF packages such as the OWASP CSRFGuard. [REF-330]
- Another example is the ESAPI Session Management control, which includes a component for CSRF. [REF-45]
- Use the “double-submitted cookie” method as described by Felten and Zeller:
- When a user visits a site, the site should generate a pseudorandom value and set it as a cookie on the user’s machine. The site should require every form submission to include this value as a form value and also as a cookie value. When a POST request is sent to the site, the request should only be considered valid if the form value and the cookie value are the same.
- Because of the same-origin policy, an attacker cannot read or modify the value stored in the cookie. To successfully submit a form on behalf of the user, the attacker would have to correctly guess the pseudorandom value. If the pseudorandom value is cryptographically strong, this will be prohibitively difficult.
- This technique requires Javascript, so it may not work for browsers that have Javascript disabled. [REF-331]
References