A security vulnerability was discovered in LiteLLM version 1.107.0, which included a credential stealer and backdoor that executed on every Python startup. This issue was identified following a previous opencode key leak. The incident also involved a series of attacks targeting AI inference providers, with one attacker gaining access to customer environment variables, leading to a pause in deployments by Railway. Users are advised to rotate their keys immediately. AI
IMPACT Compromised AI inference providers and libraries can disrupt development and deployment pipelines, necessitating immediate security measures.
RANK_REASON A specific software package (LiteLLM) was found to contain malicious code, impacting its users and related services.
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →