Popular AI gateway tool LiteLLM was compromised through supply chain attack, allowing attackers to intercept API requests and credentials

LiteLLM, a widely-used proxy for multiple AI model APIs, was exploited as a backdoor to intercept sensitive data and API keys from users

The supply chain compromise affected developers relying on LiteLLM to manage requests across platforms like OpenAI, Anthropic, and other AI providers

Attackers gained access to user credentials and API tokens, potentially allowing them to impersonate users and access their AI service accounts

The incident highlights risks in the AI infrastructure ecosystem where open-source gateway tools serve as central points of vulnerability for downstream applications