LiteLLM Malicious version 1.107.0 disclosed to be capable of stealing credentials and keys
BIBIBI
AT A GLANCE
SlowMist stated that the LiteLLM malicious version 1.107.0 can steal sensitive data and supports remote code execution.
Article
LiteLLM hit by another supply-chain attack, malicious version 1.107.0 can steal credentials and keys
PANews October 22 reported, citing SlowMist, that LiteLLM was hit by another supply-chain attack, and the malicious version [email protected] affected PyPI packages and Docker images, including floating tags such as latest and main-stable. The injected malicious code executes automatically when Python starts, stealing cloud credentials,AI API keys,SSH private keys and other sensitive data, and supporting remote code execution. The affected versions have been removed or replaced. SlowMist recommends that users avoid using the 1.107.0 version, rebuild affected environments from clean images, rotate potentially exposed credentials, and pin the version to 1.106.2。
In a October 22 message, PANews cited SlowMist as disclosing that LiteLLM was hit by another supply-chain attack.
02
The disclosure stated that the malicious version was [email protected], affecting PyPI packages and Docker images, including floating tags such as latest and main-stable.
03
The disclosure stated that the injected malicious code executes automatically when Python starts.
04
The disclosure stated that the malicious code can steal cloud credentials,AI API keys,SSH private keys and other sensitive data, and supports remote code execution.
05
The original text stated that the affected versions had been removed or replaced.
06
SlowMist recommends avoiding the 1.107.0 version, rebuilding affected environments from clean images, rotating potentially exposed credentials, and pinning the version to 1.106.2。
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
This is a software supply-chain security alert: the LiteLLM 1.107.0 version described in the report was injected with malicious code. If the relevant PyPI package was installed or an affected Docker floating tag was used, the code could run when Python starts and obtain credentials and private keys from the environment.
Why it matters to readers
Cloud credentials,AI API keys or SSH private keys, if exposed, could leave the relevant accounts, services, and hosts vulnerable to unauthorized access; remote code execution also means that attackers could run additional code.
If you use LiteLLM or have pulled relevant Docker images from tags such as latest or main-stable, you should check the version, avoid 1.107.0, and rebuild the environment and rotate potentially exposed credentials as recommended in the original text.
Risks and unknowns
The evidence did not include LiteLLM an official announcement or independent technical validation.
The evidence cannot confirm the initial release time, withdrawal time, or complete list of Docker images for the malicious version.
The evidence cannot confirm whether any credentials have already been exposed, downstream services compromised, or affected organizations disclosed.
The specific platform scope of “the affected versions have been removed or replaced” is unclear.
Related Developments
Loading event timeline…
Related concepts
supply-chain attack
This term is not in the glossary yet. Browse related concepts in the glossary.
