SAFETY & PRIVACY
How the Wukong Health Navigator keeps you safe
Plain-language explanation of the protections built into every health navigator agent — what it is, what it refuses to do, how it handles emergencies, and what we store.
An AI companion, not a doctor
The navigator synthesizes public medical literature and the published work of leading specialists into health education. It is not a licensed physician, nurse, or pharmacist, and it is never your doctor. It does not examine you, order tests, or replace a clinical visit. Use it to understand your questions better — then take what you learn to a real clinician.
Emergencies are handled first, by design
If your message describes a possible emergency — chest pain or pressure, stroke signs (face droop, arm weakness, speech trouble), severe breathing difficulty, severe or uncontrolled bleeding, fainting, or a severe allergic reaction — the agent leads with urgent action: seek emergency care now. This is not left to the model's mood. A deterministic safety check scans every message for these warning signs before the AI replies, and forces a help-first response when one is detected.
If you mention suicide or self-harm
If you express thoughts of suicide or self-harm, the agent responds first — immediately and without judgement. It will not minimize your feelings or offer hollow reassurance. It directs you to immediate help: your local emergency number, a suicide or crisis line, and reaching a trusted person right now. The same deterministic check that catches medical emergencies also catches these signals, so this response does not depend on the model getting it right on its own.
It will never diagnose or prescribe
The agent will not tell you "you have X." It frames possibilities — "this pattern can be consistent with X, Y, or Z" — and reminds you that only a clinician who examines you and reviews your tests can diagnose. It will never tell you to start, stop, raise, or lower a medication or dose. At most it explains the kinds of treatment doctors commonly consider, always paired with: your prescribing doctor decides what fits you.
Honesty over comfort
The agent's goal is to be correct, not to please. When it is uncertain, it says so. It does not invent citations or statistics — if it is unsure of a number, it rounds, hedges, or leaves it out. It will not tell you "this is definitely nothing," "you don't need to see a doctor," or "your doctor is wrong." Every interpretation it offers is paired with a concrete "when to seek care" trigger.
Your privacy
Your conversations are stored so the agent can remember context across sessions. They are scoped to your own account — row-level security at the database means you only ever see your own data, and other users cannot see yours. We do not sell your health conversations. For exactly what is collected, how long it is kept, and your rights, see the full privacy policy below.