Gradio is an open-source Python package designed for quick prototyping. This is a data validation vulnerability affecting several Gradio components, which allows arbitrary file leaks through the post-processing step. Attackers can exploit these components by crafting requests that bypass expected input constraints. This issue could lead to sensitive files being exposed to unauthorized users, especially when combined with other vulnerabilities, such as issue TOB-GRADIO-15. The components most at risk are those that return or handle file data. Vulnerable Components: 1. String to FileData: DownloadButton, Audio, ImageEditor, Video, Model3D, File, UploadButton. 2. Complex data to FileData: Chatbot, MultimodalTextbox. 3. Direct file read in preprocess: Code. 4. Dictionary converted to FileData: ParamViewer, Dataset. Exploit Scenarios: 1. A developer creates a Dropdown list that passes values to a DownloadButton. An attacker bypasses the allowed inputs, sends an arbitrary file path (like /etc/passwd), and downloads sensitive files. 2. An attacker crafts a malicious payload in a ParamViewer component, leaking sensitive files from a server through the arbitrary file leak. This issue has been resolved in gradio>5.0. Upgrading to the latest version will mitigate this vulnerability. There are no known workarounds for this vulnerability.
The product exposes sensitive information to an actor that is not explicitly authorized to have access to that information.
| Name | Vendor | Start Version | End Version |
|---|---|---|---|
| Gradio | Gradio_project | * | 5.0.0 (excluding) |
There are many different kinds of mistakes that introduce information exposures. The severity of the error can range widely, depending on the context in which the product operates, the type of sensitive information that is revealed, and the benefits it may provide to an attacker. Some kinds of sensitive information include:
Information might be sensitive to different parties, each of which may have their own expectations for whether the information should be protected. These parties include:
Information exposures can occur in different ways:
It is common practice to describe any loss of confidentiality as an “information exposure,” but this can lead to overuse of CWE-200 in CWE mapping. From the CWE perspective, loss of confidentiality is a technical impact that can arise from dozens of different weaknesses, such as insecure file permissions or out-of-bounds read. CWE-200 and its lower-level descendants are intended to cover the mistakes that occur in behaviors that explicitly manage, store, transfer, or cleanse sensitive information.