A critical SQL injection vulnerability (CVE-2026-42208) in the LiteLLM open-source AI gateway was exploited days after public disclosure to access database tables containing API keys, provider credentials, and environment configuration, Sysdig reports. Maintainers released LiteLLM 1.83.7 to fix the issue by parameterizing the key verification query, and users are advised to update immediately or disable error logs to mitigate the exploitation path. #LiteLLM #CVE-2026-42208
Keypoints
- A pre-auth SQL injection in the proxy API key verification (CVE-2026-42208) allowed unauthenticated requests to inject into database queries.
- Attackers exploited the flaw within 36 hours of the advisory being indexed and targeted tables with API keys, provider credentials, and env vars.
- Exploitation used a specially crafted Authorization header and the proxy’s error-handling path to retrieve query results.
- Observed attacks appeared automated, ran 21 minutes apart with rotated IPs, and showed no confirmed post-extraction abuse.
- LiteLLM 1.83.7 fixes the issue by passing caller-supplied values as separate parameters; update immediately or disable error logs as mitigation.
Read More: https://www.securityweek.com/fresh-litellm-vulnerability-exploited-shortly-after-disclosure/