Check Point finds two LiteLLM flaws; update to 1.83.9 or later

Check Point Research said on 13 October 2026 that it found two high risk vulnerabilities in LiteLLM, the open-source gateway many teams use to route apps to OpenAI, Google, Anthropic and other providers. CVE-2026-35029 let low privileged users, including read only ones, reach admin functions; in a controlled test Check Point read server files, overwrote admin credentials and ran outside code. CVE-2026-59823 is a server side request forgery flaw that let a user point the gateway at their own server and capture the AI provider API keys it sends. Check Point says both are fixed in LiteLLM 1.83.9 and later, after a coordinated disclosure that began in October 2025. The GitHub advisories agree: the privilege escalation advisory lists 1.83.0 as the first patched version, and the SSRF advisory lists 1.83.9. Check Point puts LiteLLM at 97 million downloads a month and over 59,000 GitHub stars, and warns it is often embedded inside other AI platforms.