Server-Side Request Forgery in haotian-liu/llava

The SSRF vulnerability in LLaVA presents a significant risk because it enables attackers to bypass network security controls and access internal resources. An attacker could leverage this vulnerability to retrieve sensitive information like AWS credentials, internal API keys, or other confidential data stored within the server's environment. Successful exploitation could lead to data breaches, unauthorized access to cloud infrastructure, and potential compromise of the entire system. The ability to make arbitrary HTTP requests effectively grants the attacker a window into the server's internal network, allowing for reconnaissance and further exploitation.

Organizations deploying LLaVA in cloud environments, particularly those utilizing AWS, are at heightened risk. Shared hosting environments where LLaVA is running alongside other applications are also vulnerable, as a successful SSRF attack could potentially compromise other tenants on the same server. Any deployment lacking robust outbound network controls is susceptible.

• python / server:

import requests
import os

def check_llava_ssrf(url):
    try:
        response = requests.get(url, timeout=5, verify=False)
        print(f"URL {url} returned status code: {response.status_code}")
        return True
    except requests.exceptions.RequestException as e:
        print(f"Error accessing {url}: {e}")
        return False

# Example usage (replace with your target URL)
target_url = os.environ.get('TARGET_URL', 'http://localhost:8000/api/some_endpoint')
check_llava_ssrf(target_url)

• generic web:

curl -v http://your-llava-instance/api/some_endpoint > /dev/null 2>&1 | grep -i 'Internal Server Error'

1. 2025-03-20Disclosure

   disclosure

1. Reserviert2024-12-02
2. Veröffentlicht2025-03-20
3. EPSS aktualisiert2026-04-02

The primary mitigation for CVE-2024-12068 is to upgrade to a patched version of LLaVA as soon as it becomes available. Until a patch is released, implement strict outbound network access controls to limit the server's ability to make external HTTP requests. This can be achieved through firewall rules, proxy configurations, or network segmentation. Consider using a Web Application Firewall (WAF) to filter outbound requests and block those targeting potentially sensitive endpoints. Regularly review and audit network configurations to ensure they adhere to security best practices.