A major supply-chain attack on the open source AI development tool LiteLLM exposed terabytes of sensitive credentials belonging to thousands of organizations, including Microsoft, Amazon, Cisco, Samsung, Salesforce, Nvidia, and others. Security firms CloudSEK and Hudson Rock reported that compromised LiteLLM versions were downloaded from the official Python Package Index and used during a roughly 40-minute window in March, allowing attackers to extract secrets from infected systems.
The stolen data reportedly includes cloud keys, repository tokens, SSH keys, Kubernetes secrets, package publishing credentials, environment variables, AI provider keys, and CI/CD pipeline secrets. In total, the firms said about 434,000 software pipelines had credentials exposed, with some data tied to more than 2,500 organizations. Many secrets were active and usable, including database passwords, third-party API keys, and cloud credentials.
Researchers traced the LiteLLM compromise to a broader campaign that also infected other software tools, including Trivy, KICS, and the Telnyx Python SDK. A group known as TeamPCP has been linked to the attack. Security experts warned that the incident shows how quickly malicious code in widely used open source dependencies can spread across the software supply chain.
Affected organizations are being urged to rotate and revoke all credentials, audit versions 1.82.7 and 1.82.8 of LiteLLM, and review logging and egress controls. The breach highlights the growing risks of rushed AI adoption combined with weak DevOps security.
Article timeline
Build this article’s timeline
Extract dated events, key actors, and open questions using the same timeline analysis available in Signal.
A major supply-chain attack on the open source AI development tool LiteLLM exposed terabytes of sensitive credentials belonging to thousands of organizations, including Microsoft, Amazon, Cisco, Samsung, Salesforce, Nvidia, and others.
🤖 AI prefill
📅WhenWhen did this happen, or what is the timeframe?
📍Where
United States and San Francisco
🤖 AI prefill
💡Why
Centers on LiteLLM, supply-chain attack, and credentials exposure
🤖 AI prefill
Login to answer and earn XP for each gap you fill.
Manipulation Score: MediumHow much this article uses emotional framing vs. straight reporting
Loaded Language3425%
Absolutist2215%
Threat/Panic7125%
Us vs Them1815%
Engagement Bait4620%
Manipulation score look right?
Factual Claims (10)
hightechnology
Terabytes worth of credentials were exposed in a supply-chain attack on LiteLLM.
hightechnology
Microsoft, Amazon, Cisco, Samsung, and Salesforce were among the entities whose access secrets were exposed.
hightechnology
CloudSEK said it found cloud keys, repository tokens, SSH keys, Kubernetes secrets, package publishing credentials, environment variables, and AI provider keys.
hightechnology
The credentials were extracted during a 40-minute window in March while victims used compromised LiteLLM versions from the official PyPI location.
mediumtechnology
Hudson Rock said it analyzed a 195TB file to make the discovery.
hightechnology
The LiteLLM compromise resulted from a previous supply-chain attack that infected Trivy.
mediumtechnology
Other software infected in the campaign includes KICS and the Telnyx Python SDK.
hightechnology
TeamPCP took credit for the attack and researchers largely corroborated the claim.
hightechnology
Some 434,000 CI/CD pipelines had credentials exposed after running the compromised LiteLLM versions.
mediumtechnology
Organizations are being urged to rotate credentials and audit environments for LiteLLM versions 1.82.7 and 1.82.8.
Comments
Login to view and post comments.