CVE-2024-32965

Lobe Chat, an open-source AI chat framework, contains an unauthenticated server-side request forgery vulnerability (CWE-918) in all versions prior to 1.19.13. The flaw stems from the JWT carried in the X-Lobe-Chat-Auth header, which stores the proxy address and OpenAI API key and can be attacker-controlled, so a request made without logging in can direct the server to issue arbitrary requests. An attacker gains the ability to scan the target's internal network, attack intranet services, and leak sensitive information such as stored API keys. Every lobe-chat deployment older than 1.19.13 is affected, with self-hosted instances reachable from untrusted networks facing the greatest risk. There is no confirmed in-the-wild exploitation or CISA KEV listing yet, but a public proof of concept exists and EPSS assigns a 27.9% probability of exploitation within 30 days (98th percentile), indicating elevated risk.

What to do: Upgrade lobe-chat to version 1.19.13 or later, as the advisory states there are no known workarounds. Until patched, avoid exposing instances directly to untrusted networks and review what internal services and credentials are reachable from the Lobe Chat host. Monitor the GitHub advisory (GHSA-2xcc-vm3f-m8rw) and prioritize patching given the high EPSS score.