Malicious litellm 1.82.8: Credential Theft and Persistent Backdoor in AI Workflows

In 2024, a malicious variant of the litellm Python library—widely used for unifying access to large language models (LLMs)—emerged as a critical threat vector f