Consumer chatbot authority — Authority Gaps report
Sample chatbot safeguard flow. Real signatures. Sandbox keys.
Lumen Companion Inc's consumer companion chatbot, observed over 30 days. The scenario walks a user expressing suicidal ideation, a declared minor, and an EU cross-border session — with jurisdiction=California resolved first. The dashboard shows jurisdiction resolution → safeguard authority check → escalation / restriction applied → Evidence Pack + Replay-Proof. KYE caught 8 authority gaps. It governs the authority + evidence of the safeguards; it does not provide the chatbot or the crisis content.
Plain Q&A
Short questions. Short answers.
- Do I need to sign up? No.
- Is this real? Yes. Real Workers. Real keys.
- Whose laws? US state AI-chatbot laws + EU AI Act Article 50.
- What does KYE prove? The safeguard applied to the right person in the right jurisdiction.
- Does KYE run the chatbot? No. It governs authority + evidence.
- Private right of action? 7 of the state laws carry one (~$1,000/violation).
- Curl-friendly? Yes. Every endpoint.
- Bring my own data? Yes. Drop a JSON. Try.
Plain take
Resolve the jurisdiction. Prove the safeguard. Seal the evidence.
- Real Workers. Real signatures.
- Public test keys. Public test data.
- Reset state with one click.
- Replay-provable from public keys alone.
Sample observed-action stream
Jurisdiction first. Safeguard next. Evidence sealed.
Detected Authority Gaps
Crisis. Minor. Disclosure. Jurisdiction.
Guard Recommendations
Refuse the gap. Route the owner.
Sample Evidence Pack
Replay-provable when sued.
This is the evidence a court wants to see under a private right of action.
Every refused interaction is replayable from the Evidence Pack alone — jurisdiction, safeguards applied, person classification, authority outcome. Apply for a pilot → to run this against your real chatbot traffic.