# Trust and human control

> People remain accountable for every real-world task. The pilot does not claim a live payment service, open registration, or an unsupervised agent. See [the trust page](https://aipaysyou.com/trust).

Agents may prepare a brief. A named human owner approves spending policy, and a moderator reviews pilot safety before a scenario is published. Evidence is scanned before a person relies on it. Malware (`CLAMAV_MALWARE_DETECTED`) is rejected. A scanner failure (`SCANNER_UNAVAILABLE` or any `ERROR`) is not a clean result. The uploader and operators see the scanner status.

Prohibited requests, reporting limits, and the current pilot boundaries are on the HTML page. This markdown file does not add permissions that the product flags have not enabled.

The agent API and the MCP server exist in code and stay disabled unless `AIPAY_AGENT_API_ENABLED` is `true` on an authenticated sandbox or production deployment. Production keeps that flag off. Read `GET /api/readiness`: `capabilities.agentApi` and `checks.agentApiEnabled` are the live answer. Payments (`AIPAY_PAYMENTS_MODE`), funding (`AIPAY_FUNDING_ENABLED`), and transfers (`AIPAY_TRANSFERS_ENABLED`) are separate flags and stay off in the no-money pilot. A page that describes a tool is not a claim that the tool is enabled.

Human-readable page: [https://aipaysyou.com/trust](https://aipaysyou.com/trust).
