AI safety prompts

Use instructions that require uncertainty to stay visible.

These copyable prompts are public guidance. They do not contact a model, enforce model behavior, expose the private Callidora鈥檚 Voice method, or replace verification and professional judgment.

01

Public boundary

The prompt guidance asks an AI to distinguish sources, dates, claims, and missing information. A prompt cannot guarantee compliance; verify every consequential output independently.

02

No private transfer

This public page does not upload a hub, send a request to Callidora鈥檚 Voice, connect a provider, or use an API credential. Copying text places it on your device clipboard.

03

Human decision

Do not use AI output as an automatic filing, complaint, accusation, diagnosis, deadline calculation, or individualized legal decision. Use current official self-help resources or qualified counsel for consequential decisions.

Bounded guidance

Tell us what happened.

You do not need legal vocabulary or the right category. Start in your own words. The handoff asks a research-capable AI to identify possible paths, show its reasons, test the record, and say what it still does not know鈥攚ithout publishing the private review method.

Public safety prompts

Use a prompt that refuses to guess.

These prompts instruct an AI not to select an official route, make a legal decision, or replace human source review. A prompt cannot enforce compliance; verify every output, source, date, and professional decision independently.

Prompt / Official source check

Use current primary and official sources for the identified jurisdiction. Link each source directly and name its publisher and publication or review date when available. If jurisdiction or authority is uncertain, stop and say what must be confirmed. Do not invent a statute, rule, filing, quote, page, or deadline.