Supply Chain Attacks and Credential Exposure Impact AI Development and Developer Platforms

Key Intelligence

• ■LiteLLM, a Python library for Large Language Models, experienced a supply chain attack that injected credential-stealing malware and a backdoor into its v1.107.0 PyPI release.
• ■The LiteLLM compromise is suspected to have originated from a GitHub token leak found in the project's AI Gateway logs, leading to the malicious package being uploaded.
• ■The attack exposed AI developer credentials and sensitive information, impacting users and downstream projects reliant on LiteLLM.
• ■Developer platform Railway separately urged all users to revoke secret keys following its own cyberattack, indicating broader credential security risks in developer tools.
• ■OpenCode AI, an AI coding tool, also exposed API keys in a suspected "distillation attack," resulting in unauthorized charges and highlighting risks in AI-specific platforms.