---
id: "CVE-2026-42208"
kind: "vulnerability"
aliases: ["CVE-2026-42208"]
title: "LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format"
severity: "critical"
cvss: 9.8
epss: null
exploited: true
status: "open"
category: "ai-tooling"
products: [{"name": "BerriAI LiteLLM", "fixed": [], "affected": "", "ecosystem": "vendor"}, {"name": "litellm", "fixed": ["1.83.7"], "affected": ">= 1.81.16, < 1.83.7", "ecosystem": "pypi"}, {"name": "litellm", "fixed": ["1.83.7"], "affected": ">= 1.81.16, < 1.83.7", "ecosystem": "vendor"}]
ecosystems: ["pypi", "vendor"]
tags: ["ai-tooling", "exploited", "litellm", "llm-gateway", "pre-auth", "python", "self-hosted", "sql-injection"]
not_affected_if: ["managed-hosting", "patched-1.83.7+"]
published_at: "2026-05-08T00:00:00+00:00"
first_seen_at: "2026-10-23T19:11:51+00:00"
updated_at: "2026-10-23T23:44:51+00:00"
sources: [{"source": "kev", "url": "https://www.cisa.gov/known-exploited-vulnerabilities-catalog?search_api_fulltext=CVE-2026-42208", "first_seen_at": "2026-10-23T19:11:51+00:00"}, {"source": "analyst", "url": null, "first_seen_at": "2026-10-23T20:45:51+00:00"}, {"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-42208", "first_seen_at": "2026-10-23T23:44:51+00:00"}]
---

# LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format

Pre-authentication SQL injection in LiteLLM Proxy's API key verification: the Authorization: Bearer value is concatenated into a SELECT on LiteLLM_VerificationToken. Any client that can reach the proxy (e.g. POST /chat/completions) can read the Postgres database, including virtual keys, the master key, stored provider credentials and proxy environment config. Exploitation in the wild was observed by Sysdig on Apr 26, 2026, and CISA added it to KEV on May 8.

_Summary, check and fix written by a model from the sources; labelled as such._

## Affected

- vendor:BerriAI LiteLLM affected ? fixed ?
- pypi:litellm affected >= 1.81.16, < 1.83.7 fixed 1.83.7
- vendor:litellm affected >= 1.81.16, < 1.83.7 fixed 1.83.7

## How to check

Run 'pip show litellm' or check the image tag: versions 1.81.16 through 1.83.6 are vulnerable. Applies only if the proxy uses a Postgres database. Review Postgres query history with BerriAI's helper query (https://gist.github.com/ishaan-berri/6f31e56e878338eb4c01990bd08378ab). Grep access logs for Authorization headers containing a single quote, UNION SELECT, or litellm_verificationtoken/litellm_credentials, especially user-agent Python/3.12 aiohttp/3.9.1 from 65.111.27.132 and 65.111.25.67.

## Fix

Upgrade LiteLLM to v1.83.10-stable (recommended), or at minimum v1.83.7. If the proxy was reachable from an untrusted network on an affected version, rotate the master key, all virtual keys and every provider credential stored in LiteLLM (OpenAI, Anthropic, Bedrock etc.), and the DB password. No config-only workaround was stated by the vendor; restricting network access to the proxy limits exposure.

## Timeline

- 2026-10-23T19:11:51+00:00 kev https://www.cisa.gov/known-exploited-vulnerabilities-catalog?search_api_fulltext=CVE-2026-42208
- 2026-10-23T20:45:51+00:00 analyst — BerriAI blog (GHSA-r75f-5x8p-qvmc) gives affected v1.81.16-v1.83.6, fixed v1.83.7+, recommends v1.83.10-stable; Sysdig confirms pre-auth SQLi in the Authorization: Bearer key check and in-the-wild attempts from Apr 26. No public PoC link read; Sysdig lists payload shapes. Advisory page itself not read directly.
- 2026-10-23T23:44:51+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-42208

## Documents

- [advisory] BerriAI: Security Update CVE-2026-42208 in LiteLLM Proxy https://docs.litellm.ai/blog/cve-2026-42208-litellm-proxy-sql-injection
- [describes] Sysdig: Targeted SQL injection against LiteLLM auth path https://sysdig.com/blog/cve-2026-42208-targeted-sql-injection-against-litellms-authentication-path-discovered-36-hours-following-vulnerability-disclosure

## From the sources

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From version 1.81.16 to before version 1.83.7, a database query used during proxy API key checks mixed the caller-supplied key value into the query text instead of passing it as a separate parameter. An unauthenticated attacker could send a specially crafted Authorization header to any LLM API route (for example POST /chat/completions) and reach this query through the proxy's error-handling path. An attacker could read data from the proxy's database and may be able to modify it, leading to unauthorised access to the proxy and the credentials it manages. This issue has been patched in version 1.83.7.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_