threatdeck
Live 14:45 UTC ·

AI Stack Alert High For CISO, SOC / IR, AppSec, AI Eng AI-assisted

LiteLLM < 1.95.0: authenticated file read via vertex_ai_credentials alias

LiteLLM proxy versions before 1.95.0 let an authenticated user read arbitrary local files through /utils/transform_request by sending a crafted vertex_ai_credentials value. The request-body blocklist covered vertex_credentials but missed this alias, so the value reached the Vertex AI credential loader. Fixed in 1.95.0.

Why it matters

A user with only the internal_user_viewer role, which can be obtained through SSO on many deployments, can read files the proxy process can read. Reading /proc/self/environ can expose LITELLM_MASTER_KEY and provider API keys, enabling full proxy-admin control.

Do this

  1. Upgrade LiteLLM to 1.95.0 or later.
  2. Until upgraded, review which users hold the internal_user_viewer role and any SSO auto-provisioning that grants it.
  3. Rotate LITELLM_MASTER_KEY and provider API keys if an affected version was exposed to untrusted users, since file reads may have exposed them.
  4. Review proxy logs for requests to /utils/transform_request that include vertex_ai_credentials.

Signals

  • NoCISA KEV
  • 8.1CVSS
  • 37/100Priority
  • 85%Confidence

Evidence

“The request-body safety blocklist rejected `vertex_credentials` but missed its alias `vertex_ai_credentials`”

GitHub Security Advisories (AI packages)

“Upgrade to `1.95.0` or later. The fix adds the missing alias to the request-body blocklist”

GitHub Security Advisories (AI packages)

Sources

Affected / related

LiteLLM proxy, arbitrary local file read

Published 2026-10-27 15:32 UTC · updated 2026-11-08 02:40 UTC · AI-assisted summary, checked against quoted sources · How we score

Following

Follow the vendors, products, AI tools, techniques and actors you care about. Matching stories, CVEs and exploited bugs are marked, a Following tab appears in the feed, and returning visits show what changed. Saved only in this browser.

Presets

Get updates

A personal RSS feed of new stories on everything you follow. Paste it into any feed reader, Slack or Teams RSS app.