Skip to content
Submit incident
AII-1002-47681
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.

SSRF vulnerability in Gradio /queue/join endpoint (CVE-2024-47167)

Gradio's /queue/join endpoint contained a Server-Side Request Forgery vulnerability in versions prior to 5.0 that could allow attackers to send HTTP requests to internal networks or upload malicious files. The vulnerability was patched in Gradio 5.0.

10 October 2024 Β· Gradio Β· Security and cyberattacks
10 October 2024
AII-1002-476812 sources ↓
Quick facts
Organization
Gradio (subject)
Category
Security and cyberattacks
Source
github.com
Date
10 October 2024
Drafted by
Raidu Incident Detection

What happened

A Server-Side Request Forgery (SSRF) vulnerability was discovered in the /queue/join endpoint of Gradio versions prior to 5.0. The vulnerability exists in the async_save_url_to_cache function, which allows attackers to supply user-controlled URLs that the server will fetch.

The vulnerability affects Gradio components like the Video component that involve URL fetching. Content from attacker-supplied URLs is stored locally on the server, potentially enabling upload of malicious files.

Reported impact

Attackers could force the Gradio server to send HTTP requests to internal servers or services within a local network, potentially leading to data exfiltration or unwanted internal requests. The vulnerability impacts users deploying Gradio servers with URL-fetching components.

Index analysis

The vulnerability was assigned CVE-2024-47167 and GHSA-576c-3j53-r9jj, with a severity rating of High. Gradio version 5.0 and later contain patches for this issue.

Governance controls

As a workaround, users can disable or restrict URL-based inputs to trusted domains only, implement allowlist-based URL validation, and ensure local or internal network addresses cannot be requested via the /queue/join endpoint.

Classification

Organization
Gradio (subject)
Category
Security and cyberattacks
Harm status
Potential harm
Intent
Unintentional
Timing
Post-deployment
AI systems
Gradio

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

github.com
https://github.com/gradio-app/gradio/security/advisories/GHSA-576c-3j53-r9jj
cve.org
NVD entry