A flaw in ChatGPTNextWeb’s NextChat component, present in versions up to 2.16.1, lets a remote attacker manipulate the API endpoint to establish an overly permissive cross‑domain policy. The vulnerability allows the component to accept requests from arbitrary, untrusted domains, potentially exposing sensitive application data and internal interactions. The weakness is identified by CWE‑346 and CWE‑942, indicating improper handling of trust for incoming content and possible path traversal via URLs.

ChatGPTNextWeb’s NextChat component, all releases up to and including v2.16.1, is affected. The flaw lies within an unnamed file of the Next.js API endpoint. No other vendors or products are referenced.

The vulnerability carries a CVSS score of 5.3, indicating a moderate severity level. EPSS information is not available, and the issue is not listed in CISA’s KEV catalog. However, a published exploit exists and can be launched remotely, so the risk of active exploitation is real. Attackers can craft requests that bypass origin restrictions through the compromised endpoint, potentially accessing internal resources.