Responsible AI controls

High-risk actions remain owner-approved through admin gates, kill-switch policy and capability boundaries.

The anti-dark-pattern pledge forbids fake scarcity, forced accounts, drip pricing and retention traps.

Pricing experiments publish public aggregate metrics at /transparency.

Personal data is limited to activation, receipts, support and delivery records.

Agent-to-agent checkout uses signed receipts and endpoint-scoped capability credentials.

Temporal product memory records deploy identity, risk and rollback-ready status.