Executive brief
FreeGPT WebUI is a web interface that provides access to AI chat completion services. A vulnerability in the authentication mechanism allows attackers to bypass required authentication checks and access protected AI providers (such as Bard) without credentials by simply omitting a provider argument in their requests.
Technical details
This is a missing authentication vulnerability (CWE-306) in the ChatCompletion.create method of g4f/__init__.py. The vulnerability occurs because the authentication check (`if provider and provider.needs_auth and not auth`) only validates credentials when a provider is explicitly specified. When no provider argument is passed, the code resolves the provider from `model.best_provider` but skips the authentication check entirely, allowing any model whose best_provider requires authentication (e.g., Model.palm → Provider.Bard with needs_auth=True) to be invoked without credentials. The backend API endpoint in server/backend.py never passes explicit provider or auth arguments, making all auth-required providers accessible unauthenticated through the web API. Attack is network-accessible and requires no prior authentication from the attacker.
Affected products
- ramon-victor freegpt-webui up to 098db3dfeb41555c2ca9269df0f13e10ec1c35dc
Timeline
- 2026-07-18: disclosed: Advisory published on GitHub Gist
- 2026-09-04: advisory: CVE-2026-85701 published
- 2026-09-04: other: Product noted as archived and no longer supported by maintainer