Remote Code Execution in LiteLLM Proxy Server

Executive Summary

This vulnerability affects LiteLLM, a proxy server that acts as an AI Gateway to call large language model (LLM) APIs in OpenAI or native format. The issue existed in versions prior to 1.84.0 and was fixed in version 1.84.0.

The vulnerability is severe, with a CVSS v4.0 base score of 9.5, indicating it is highly exploitable and impactful.

Useful 0 Not Useful 0

Impact Analysis

Given the high CVSS score of 9.5, this vulnerability can have a critical impact. It likely allows an attacker to exploit the LiteLLM proxy server remotely without privileges or user interaction, potentially leading to severe consequences such as complete compromise of the system or data.

Useful 0 Not Useful 0

Mitigation Strategies

The vulnerability in LiteLLM is fixed in version 1.84.0. To mitigate this vulnerability, you should immediately upgrade your LiteLLM installation to version 1.84.0 or later.

Useful 0 Not Useful 0

Compliance Impact

The vulnerability in LiteLLM allows unauthenticated access to protected management routes due to an authentication bypass via Host-header manipulation. This critical flaw can lead to unauthorized access, impacting confidentiality, integrity, and availability of the system.

Such unauthorized access and potential data exposure could negatively affect compliance with common standards and regulations like GDPR and HIPAA, which require strict controls on access to sensitive data and systems.

However, the vulnerability is mitigated in many deployments by upstream components that validate or normalize the Host header, and it is fully patched in version 1.84.0.

Useful 0 Not Useful 0

Detection Guidance

CVE-2026-49468 is an authentication bypass vulnerability in LiteLLM proxy versions prior to 1.84.0 caused by a Host-header parsing issue. Detection involves monitoring for suspicious or crafted Host headers that attempt to manipulate authentication by bypassing the auth gate.

Since the vulnerability exploits the Host header to bypass authentication, you can detect attempts by capturing and inspecting HTTP requests to the LiteLLM proxy for unusual or unexpected Host header values.

Suggested commands to detect potential exploitation attempts include using network packet capture tools like tcpdump or tshark to filter HTTP traffic to the proxy and grep for suspicious Host headers.

• tcpdump -i <interface> -A -s 0 'tcp port <proxy_port>' | grep -i 'Host:'
• tshark -i <interface> -Y 'http.host' -T fields -e http.host

Additionally, reviewing proxy logs for requests with unexpected Host header values or authentication bypass patterns can help identify exploitation attempts.

If upgrading to version 1.84.0 is not immediately possible, placing the proxy behind an upstream component that validates or normalizes the Host header (such as a CDN, WAF, or reverse proxy with allowlists) can help mitigate the risk and assist in detection.

Useful 0 Not Useful 0