Summary: Route model requests through an application backend instead of placing a shared API credential in public client code.
A browser or mobile application cannot reliably protect a shared API key. Users can inspect bundles, runtime requests, storage, and application packages.
A frontend environment variable is not automatically private. Build tools commonly place variables intended for client use into downloadable code.
Send the userβs authenticated request to an application backend instead. The backend can validate input, apply feature policy, read the credential from server-side secret storage, and call the model endpoint.
VectorNode is our product. Its public page describes one-key access to GPT, Claude, Gemini, DeepSeek, Qwen, Midjourney, Kling, and other models, together with unified credits, transparent pricing, and usage logs. It can be connected behind a server-owned boundary: https://www.vectronode.com
Disclosure: VectorNode is our product, and this link is included for readers who want to explore it.
Avoid building a blind relay. The backend should authenticate the application user, constrain accepted request shapes, control which product features can invoke which model operations, and remove authorization data from logs.
If a credential appears in public code or logs, treat it as exposed and follow the relevant revocation or rotation process.
A backend proxy is one security control, not a complete security program. Its value is giving the application a place to enforce policy before the model call.
For further actions, you may consider blocking this person and/or reporting abuse
Top comments (0)