FastGPT versions up to 4.14.11 contain an SSRF vulnerability due to insufficient validation in the isInternalAddress() function. This function blocks cloud metadata endpoints by checking if the full URL starts with certain hardcoded strings. However, at least seven different URL encoding methods can bypass this check, allowing requests to reach the cloud metadata service. The intended broader internal IP checks are disabled by default because the CHECK_INTERNAL_IP flag is false, meaning no additional filtering occurs. This results in SSRF requests potentially accessing sensitive internal metadata endpoints. No official remediation or patch has been released as of the publication date.

An attacker with at least low-level privileges (PR:L) can exploit this SSRF vulnerability remotely (AV:N) without user interaction (UI:N) to access cloud metadata services, potentially exposing sensitive information (C:H). The vulnerability does not impact integrity or availability but can lead to significant confidentiality breaches. The scope is changed (S:C), indicating that the vulnerability affects resources beyond the vulnerable component itself.

At the time of publication, no official patch or remediation is available. Users should monitor the vendor's advisory channels for updates. As a temporary measure, enabling and enforcing the broader internal IP checks (CHECK_INTERNAL_IP set to true) may reduce risk by blocking requests to private IP ranges, though this is not confirmed as a complete mitigation. Avoid exposing FastGPT instances to untrusted networks until a fix is released.