Method visibility
Rules-based routing, browser calculations, on-device models and server AI are labelled separately. A fallback is not presented as the requested provider result.
Responsible AI · reviewed 2026-08-10
Useful, inspectable and never silently authorized.
FTN identifies the method and keeps generated output in the state it has actually reached.
Source discipline
Current factual claims require sources and dates. AI output without verification remains a draft, suggestion or generated preview.
Action consent
Generated content is not submitted, emailed, posted, purchased or published until the user approves the named destination and action.
Private data
Civic evidence, Love activity, private ibis content and private media are not silently reused for advertising, model training or unrelated recommendations.
Provider control
Only FTN-owned services or deliberately connected, revocable provider scopes may receive data. Passwords, browser sessions and unconnected consumer accounts are never repurposed.
Human escalation
Safety, legal, moderation, regulated and high-impact decisions require an accountable human path. AI confidence is not identity verification or professional advice.