Start here
The free tool below runs entirely in this browser — no account, nothing stored or transmitted. Redact anything you would not want kept before you type it.
Open the companion tool →Why fluency is not accuracy
A model's confidence has no relationship to its correctness — it sounds equally sure when it is right and when it is inventing. The only reliable signal is the subject: medical and legal questions carry stakes that no general model is licensed or accountable for.
The check is a routing rule: for health and law, the answer is a starting point to verify with a qualified human, never the thing you act on. For everything else, verify the claim at a primary source before it becomes a decision.
The protocol
- Describe the answer and what you were about to do with it.
- Read the trust basis: the subject and the stakes.
- For medical or legal questions, verify with a qualified human before acting.
- For everything else, check the claim at a primary source.
What the tool does
Describe an AI answer you were about to act on and get a trust reading by subject and stakes. It runs in this browser; nothing is sent or stored.
Limit first: This is a routing aid, not medical or legal advice. For health and legal questions, a qualified human is the only safe place to act. For urgent symptoms or crisis, use 911 or 988 now.
Open the companion tool →One human next step
For medical questions, talk to a clinician; for legal questions, a qualified professional. For urgent symptoms or a crisis, use 911 or 988, not a chatbot.
Research log and safety checks
AI-assist path: not applicable — the tool's job is to route the reader to a human for exactly the subjects where another AI layer would be the problem. §6.10: not a frontier-pace topic.
§0.5 protection result: category is informed-consent-literacy; crisis-adjacent: no; fear/urgency/scarcity toolkit used: no. The protective function is free and the paid feature is convenience only.