Threat
When LLM Gateways Go Wrong: Findings from LiteLLM - Check Point Research
First reported · Discovered checkpoint.com
Page published
Earliest dated coverage: 15 Oct 2026 · First observed: 18 Oct 2026 · Latest dated coverage: 15 Oct 2026
Coverage timeline
Single-source incident — one report is available.
Why it matters
LiteLLM gateways centralize provider credentials and request routing, so defenders should patch standalone and embedded deployments and rotate exposed keys if compromise is suspected.
Check Point Research demonstrated two LiteLLM vulnerabilities against version 1.77.7: CVE-2026-35029 allowed low-privileged authenticated users to modify gateway configuration, enabling arbitrary file reads, administrator takeover, and remote code execution. CVE-2026-59823 enabled SSRF through the Clientside LLM Credentials feature to redirect provider requests and exfiltrate API keys. Both vulnerabilities are fixed in LiteLLM 1.83.9 and later; the supplied reporting describes controlled demonstrations, not observed malicious exploitation.