---
id: CVE-2026-114612
title: >-
  LiteLLM: Privilege Escalation to Proxy Admin via Cross-Domain Reuse of the
  Salt Key
summary: >-
  LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native)
  format. From 1.87.0 until 1.100.4, 1.101.3, 1.102.2, 1.103.1, and
  1.104.0-rc.2, the proxy reuses one cryptographic context for at-rest secrets
  and session toke…
severity: critical
cvss: 9.9
cvssVector: 'CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H'
cvssSource: cna
cwe:
  - CWE-269
  - CWE-345
  - CWE-441
vendor: BerriAI
product: litellm
affected:
  - 'litellm >= 1.87.0, < 1.100.4'
  - 'litellm >= 1.101.0, < 1.101.3'
  - 'litellm >= 1.102.0, < 1.102.2'
  - 'litellm >= 1.103.0, < 1.103.1'
  - 'litellm >= 1.104.0-rc.1, < 1.104.0-rc.2'
published: '2026-10-22'
updated: '2026-10-22'
sourceUpdated: '2026-10-22T18:01:52.768Z'
source: CVEORG
sourceUrl: 'https://www.cve.org/CVERecord?id=CVE-2026-114612'
references:
  - url: 'https://github.com/BerriAI/litellm/security/advisories/GHSA-7hp6-4w63-5g45'
    label: 'https://github.com/BerriAI/litellm/security/advisories/GHSA-7hp6-4w63-5g45'
  - url: >-
      https://github.com/BerriAI/litellm/commit/1d5fdc87fed66b49e1df8cf7d708ad93249eb7d0
    label: >-
      https://github.com/BerriAI/litellm/commit/1d5fdc87fed66b49e1df8cf7d708ad93249eb7d0
  - url: >-
      https://github.com/BerriAI/litellm/commit/9161d7f9d6073603ba61ee4aa5e2c0ef5b37a81b
    label: >-
      https://github.com/BerriAI/litellm/commit/9161d7f9d6073603ba61ee4aa5e2c0ef5b37a81b
  - url: >-
      https://github.com/BerriAI/litellm/commit/9fa1a641119dd0d4fe43e93622eae5f482ceb63f
    label: >-
      https://github.com/BerriAI/litellm/commit/9fa1a641119dd0d4fe43e93622eae5f482ceb63f
  - url: >-
      https://github.com/BerriAI/litellm/commit/a80eacacee118fbed54e3d98f239d17a63fac818
    label: >-
      https://github.com/BerriAI/litellm/commit/a80eacacee118fbed54e3d98f239d17a63fac818
  - url: >-
      https://github.com/BerriAI/litellm/commit/f7cd90f09e9716652b38daec9f0640cdf49d582b
    label: >-
      https://github.com/BerriAI/litellm/commit/f7cd90f09e9716652b38daec9f0640cdf49d582b
  - url: 'https://github.com/BerriAI/litellm/releases/tag/v1.100.4'
    label: 'https://github.com/BerriAI/litellm/releases/tag/v1.100.4'
  - url: 'https://github.com/BerriAI/litellm/releases/tag/v1.101.3'
    label: 'https://github.com/BerriAI/litellm/releases/tag/v1.101.3'
  - url: 'https://github.com/BerriAI/litellm/releases/tag/v1.102.2'
    label: 'https://github.com/BerriAI/litellm/releases/tag/v1.102.2'
  - url: 'https://github.com/BerriAI/litellm/releases/tag/v1.103.1'
    label: 'https://github.com/BerriAI/litellm/releases/tag/v1.103.1'
  - url: 'https://github.com/BerriAI/litellm/releases/tag/v1.104.0-rc.2'
    label: 'https://github.com/BerriAI/litellm/releases/tag/v1.104.0-rc.2'
tags:
  - cve.org
ingestedAt: '2026-10-22T22:00:50.361Z'
---

## Overview

LiteLLM is a proxy server (AI Gateway) to call LLM APIs in OpenAI (or native) format. From 1.87.0 until 1.100.4, 1.101.3, 1.102.2, 1.103.1, and 1.104.0-rc.2, the proxy reuses one cryptographic context for at-rest secrets and session tokens, allowing an authenticated internal user to have crafted metadata containing a forged administrator credential encrypted and returned by the API-key flow. Replaying that ciphertext as a bearer token causes the proxy to trust the forged proxy administrator identity, permitting administrative access and arbitrary commands through the MCP stdio endpoint. Versions 1.87.0 through 1.90.x require EXPERIMENTAL_UI_LOGIN to be enabled, while versions after 1.91.0 are exploitable in the default configuration. This issue was fixed in LiteLLM 1.100.4, 1.101.3, 1.102.2, 1.103.1, and 1.104.0-rc.2.

## Affected

- `litellm >= 1.87.0, < 1.100.4`
- `litellm >= 1.101.0, < 1.101.3`
- `litellm >= 1.102.0, < 1.102.2`
- `litellm >= 1.103.0, < 1.103.1`
- `litellm >= 1.104.0-rc.1, < 1.104.0-rc.2`

## Remediation

Refer to the linked advisories for vendor-supplied fixes and affected version ranges.