CVE-2026-42208
About
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.
Rainforest analyst review
The vulnerability is a SQL injection in the LiteLLM proxy, present from 1.81.16 before 1.83.7. During the proxy's API-key check, a database query concatenated the caller-supplied key value into the query text instead of binding it as a separate parameter. An unauthenticated attacker sends a crafted Authorization header to any LLM API route, such as POST /chat/completions, and the value reaches the vulnerable query through the proxy's error-handling path. From there the attacker can read from the proxy's database and may be able to modify it, exposing the proxy and the credentials it stores.
LiteLLM is an AI gateway that fronts multiple LLM providers, so its database is precisely where upstream provider API keys, virtual keys, and usage data live; reading or altering it hands an attacker the credentials the gateway exists to manage and potentially control over access to the models behind it. Because the trigger is an unauthenticated header on ordinary API routes, any internet-reachable proxy is exposed. Upgrade to 1.83.7 or later, and given the credential exposure, rotate the provider API keys and virtual keys held by the proxy and review its database and logs for tampering.
References
- https://github.com/BerriAI/litellm/releases/tag/v1.83.7-stable
- https://github.com/BerriAI/litellm/security/advisories/GHSA-r75f-5x8p-qvmc
- https://access.redhat.com/security/cve/CVE-2026-42208
- https://bugzilla.redhat.com/show_bug.cgi?id=2463965
- https://security.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-42208.json
- https://www.cisa.gov/known-exploited-vulnerabilities-catalog?field_cve=CVE-2026-42208
Related CVEs
Frequently asked questions
What is CVE-2026-42208?
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.
How severe is CVE-2026-42208?
CVE-2026-42208 carries a CVSS 3.1 base score of 9.8, rated critical. On the Rainforest Risk Index — which weighs the severity alongside how easy the flaw is to reach and how broad its impact is — it scores 99 out of 100, in the critical band.
How is CVE-2026-42208 exploited?
According to the CVSS vector (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H): attack vector Network, attack complexity Low, privileges required None, user interaction None. Impact on confidentiality High, integrity High and availability High.
Which products are affected by CVE-2026-42208?
Public advisories list the following as affected: litellm. Check the references on this page for the exact versions each vendor confirms.
How do I fix CVE-2026-42208?
Apply the fix the vendor published — the references on this page link to the primary advisories and patches. When patching can't happen right away, reduce the exposure of the affected component and watch it for exploitation attempts. Rainforest customers see this vulnerability correlated to their own assets and prioritized by real exposure, not by score alone.
Empower Your Security Strategy with Rainforest
Discover vulnerabilities early, prioritize critical threats, and protect what truly matters. Rainforest streamlines your security operations, saving you time and reducing costs, so you can focus on what drives your business forward.