Skip to content
NewA calmer, faster ZeroHour

LiteLLM 1.107.0 release carried credential-stealing malware

Original title: LiteLLM Supply Chain Compromise Exposes AI Developer Credentials to Malware Attacks

GBHackersby Divya ()2 min read1 · 1 read
Part of a story covered by 16 sourcesTrojanized LiteLLM 1.107.0 removed; users urged to rotate secretsMerged summary & timeline
AI summary · x-ai/grok-4.7Summaries by Vani AI

Malicious Python startup hooks can collect cloud and AI credentials even when applications never import LiteLLM.

Socket reported on October 21, 2026 that LiteLLM 1.107.0 PyPI wheels and container images installed a credential collector. Startup files can run when Python or pip starts, targeting cloud, SSH, browser, and AI assistant secrets, then uploading encrypted JSON. Container tags v1.107.0, latest, main-latest, and main-stable resolved to the same image. Socket found no evidence of successful theft from individual deployments, and how a repository override was authorized remains unclear.

Unconfirmed: How the source-owner override was authorized is unknown, and no successful theft from a deployment was shown.

Key facts
  • LiteLLM receives about 23 million weekly downloads, Socket said.
  • Seven malicious wheels were flagged about seven minutes after publication.
  • The collector covers AWS, Azure, Google Cloud, Kubernetes, SSH keys, and assistants including Codex, Claude, Gemini, and Cursor.
  • A separate loader can fetch unsigned Python from a GitHub pointer and retries hourly.
  • Build provenance named Berri-AI/litellm rather than BerriAI/litellm.
Security assessment · AI, from the article

Impact: Steals cloud, SSH, browser, and AI assistant credentials

ProductAffected
BerriAI LiteLLM1.107.0
What to do
  1. 1.Block litellm 1.107.0, inspect image digests, and isolate hosts where Python ran after install.
  2. 2.Rotate potentially exposed credentials from a clean system and rebuild from reviewed artifacts.
  3. 3.Do not rely on removing LiteLLM alone to stop an existing loader.
Analyze this story with Vani AIImpact, who is affected and what defenders should do, in a new chat

Indicators of compromise

auto-extracted · verify before use · export allAll
TypeIndicatorContext
domainraw.githubusercontent.comxp://www[.]litellm[.]dev/healthcheck Loader pointer hxxps://raw[.]githubusercontent[.]com/letters9813/pythonclient-test/main/domain.txt Upload he
domainwww.litellm.devType Indicator Package litellm==1.107.0 Exfiltration hxxp://www[.]litellm[.]dev/healthcheck Loader pointer hxxps://raw[.]githubusercont
sha25634bef04fe88ac9d0aae0475630b1738681d15b94f233ac8aa46eacf5968305fabe93139cc4ebef3ef9e2efe9531390be25 sitecustomize.py SHA-256 34bef04fe88ac9d0aae0475630b1738681d15b94f233ac8aa46eacf5968305fa Note: IP addresses and domains are intentionally defanged (
sha2569dd0a3926f96a37eb86cf1ee2310d9be93139cc4ebef3ef9e2efe9531390be25n marker /tmp/.pc_once_72129118c59f5636c0bb cook.py SHA-256 9dd0a3926f96a37eb86cf1ee2310d9be93139cc4ebef3ef9e2efe9531390be25 sitecustomize.py SHA-256 34bef04fe88ac9d0aae0475630b1738681
urlhttps://raw.githubusercontent.com/letters9813/pythonclient-test/main/domain.txtation hxxp://www[.]litellm[.]dev/healthcheck Loader pointer hxxps://raw[.]githubusercontent[.]com/letters9813/pythonclient-test/main/domain.txt Upload header X-Token: XWiA8gZNa7NcX2XRoAEb Loader lock
urlhttp://www.litellm.dev/healthcheckn. IoC Type Indicator Package litellm==1.107.0 Exfiltration hxxp://www[.]litellm[.]dev/healthcheck Loader pointer hxxps://raw[.]githubusercontent[.]com/le

From the original article

2 min read at gbhackers.com

LiteLLM version 1. 107. 0 was compromised in a supply chain attack that distributed credential-stealing malware through PyPI wheels and container images.

GBHackers

By Divya · gbhackers.com

Read the full article at GBHackers

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.