CVE-2024-2206: Gradio: SSRF exposes internal HuggingFace endpoints

MEDIUMPoC AVAILABLECISA: TRACK*

Published March 27, 2024

CISO Take

Any Gradio deployment on Hugging Face Spaces is exploitable without credentials — an attacker sends a crafted X-Direct-Url header to inject arbitrary URLs into the proxy allowlist, then pivots to internal endpoints including cloud metadata services. Patch immediately to commit 49d9c48 or disable the /proxy route if unused. Treat any unpatched Gradio instance as a potential pivot point into your ML hosting infrastructure.

What is the risk?

Effective risk exceeds the CVSS 6.5 score for ML teams. Zero prerequisites (no auth, no user interaction, network-accessible) make exploitation trivial for any attacker who identifies a Gradio instance. In Hugging Face Spaces, SSRF can reach the instance metadata service (169.254.169.254), exposing cloud credentials and enabling lateral movement. The attack surface is large — Gradio is the dominant ML demo framework with tens of thousands of public deployments.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
Gradio	pip	—	No patch
43.9KOpenSSF 5.9735 dependentsPushed 6d ago29% patched~108d to patchFull package profile →

Do you use Gradio? You're affected.

How severe is it?

CVSS 3.1

6.5 / 10

EPSS

0.4%

chance of exploitation in 30 days

Higher than 37% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○CISA SSVC: Public PoC

○Public PoC indexed (trickest/cve)

Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AVNetwork

ACLow

PRNone

UINone

SUnchanged

CLow

ILow

ANone

What should I do?

5 steps

PATCH
Update to the version containing commit 49d9c48537aa706bf72628e3640389470138bdc6 — verify via pip show gradio.
WORKAROUND
If patching is blocked, disable or firewall the /proxy route at the reverse proxy layer.
NETWORK
Block egress to 169.254.169.254 (IMDSv1) from Gradio containers; enforce IMDSv2 with hop-limit=1 on AWS/GCP/Azure.
DETECTION
Alert on HTTP requests containing X-Direct-Url headers in your WAF or ingress logs.
AUDIT
Review all Gradio versions in use across ML serving infrastructure; treat any pre-patch version as compromised if internet-facing.

What does CISA's SSVC say?

DecisionTrack*

Exploitationpoc

AutomatableYes

Technical Impactpartial

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Data ExtractionAuth BypassFrameworkInferenceAPIAML.T0006- Active Scanning AML.T0025- Exfiltration via Cyber Means AML.T0040- AI Model Inference API Access AML.T0049- Exploit Public-Facing Application

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15-Accuracy, robustness and cybersecurity

ISO 42001

6.1.2-AI risk assessment8.4-AI system operation

NIST AI RMF

GOVERN 1.2-Policies and procedures for AI risk governanceMANAGE 2.2-Mechanisms for vulnerability and incident response

OWASP LLM Top 10

LLM05:2025-Improper Output Handling

Frequently Asked Questions

What is CVE-2024-2206?

Is CVE-2024-2206 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2024-2206, increasing the risk of exploitation.

How to fix CVE-2024-2206?

1. PATCH: Update to the version containing commit 49d9c48537aa706bf72628e3640389470138bdc6 — verify via pip show gradio. 2. WORKAROUND: If patching is blocked, disable or firewall the /proxy route at the reverse proxy layer. 3. NETWORK: Block egress to 169.254.169.254 (IMDSv1) from Gradio containers; enforce IMDSv2 with hop-limit=1 on AWS/GCP/Azure. 4. DETECTION: Alert on HTTP requests containing X-Direct-Url headers in your WAF or ingress logs. 5. AUDIT: Review all Gradio versions in use across ML serving infrastructure; treat any pre-patch version as compromised if internet-facing.

What systems are affected by CVE-2024-2206?

This vulnerability affects the following AI/ML architecture patterns: model serving, ML demo infrastructure, inference endpoints, agent frameworks, RAG pipelines.

What is the CVSS score for CVE-2024-2206?

CVE-2024-2206 has a CVSS v3.1 base score of 6.5 (MEDIUM). The EPSS exploitation probability is 0.42%.

What is the AI security impact?

Affected AI Architectures

model servingML demo infrastructureinference endpointsagent frameworksRAG pipelines

MITRE ATLAS Techniques

AML.T0006Active Scanning

AML.T0025Exfiltration via Cyber Means

AML.T0040AI Model Inference API Access

AML.T0049Exploit Public-Facing Application

Compliance Controls Affected

EU AI Act: Article 15

ISO 42001: 6.1.2, 8.4

NIST AI RMF: GOVERN 1.2, MANAGE 2.2

OWASP LLM Top 10: LLM05:2025

What are the technical details?

Original Advisory

An SSRF vulnerability exists in the gradio-app/gradio due to insufficient validation of user-supplied URLs in the `/proxy` route. Attackers can exploit this vulnerability by manipulating the `self.replica_urls` set through the `X-Direct-Url` header in requests to the `/` and `/config` routes, allowing the addition of arbitrary URLs for proxying. This flaw enables unauthorized proxying of requests and potential access to internal endpoints within the Hugging Face space. The issue arises from the application's inadequate checking of safe URLs in the `build_proxy_request` function.

Exploitation Scenario

Attacker scans Hugging Face Spaces for Gradio instances (trivial via HF API). Sends GET / with header X-Direct-Url: http://169.254.169.254/latest/meta-data/iam/security-credentials/ — this injects the metadata URL into self.replica_urls. Follow-up request to /proxy retrieves live AWS IAM credentials. Attacker uses exfiltrated credentials to access S3 buckets containing training data, model weights, or downstream services. Full chain: unauthenticated HTTP request → cloud credential exfiltration → data breach or model IP theft.

Weaknesses (CWE)

CWE-918Server-Side Request Forgery

CWE-918 — Server-Side Request Forgery: The web server receives a URL or similar request from an upstream component and retrieves the contents of this URL, but it does not sufficiently ensure that the request is being sent to the expected destination.

[Implementation, Operation] If there is a known list of applications, IP, hostname, and/or domains to which the product is allowed to communicate, then use allowlists to ensure that outgoing requests are restricted to these known entities. The allowlists could be in the application code or possibly enforced by firewalls or network segmentation [REF-1511].
[Implementation, Operation] If the product is expected to be able to connect to arbitrary IP addresses or domains, consider using a denylist that includes private network addresses, internal domain names, etc. The denylist could be in the application code or possibly enforced by firewalls or network segmentation [REF-1511].

Source: MITRE CWE corpus.