← Back to AIA

Product FAQ privacy

This chat explains AIA and can demonstrate its general workflow using a brief situation you voluntarily describe. Please keep personal case details, names, account numbers and documents out of it.

Your questions are temporary

The conversation is held in this page’s memory. It is cleared when you select Clear chat or leave the page. It is not transferred into an AIA account. The FAQ does not write conversation content to a database, browser storage, analytics events or application logs.

AI processing

Suggested questions use approved product answers without an AI request. For free-text AI processing, the server sends your questions or brief description and AIA’s approved product context directly to OpenAI’s Luna model. It selects approved product information and may add one brief AI-written acknowledgment of a current personal situation, limited to 200 characters and one sentence. The main explanation stays approved AIA copy. An introduction that fails the server’s checks is omitted; those checks cannot guarantee that every incorrect or inappropriate sentence is detected. Product and hypothetical questions do not call for a personal acknowledgment. This homepage does not research your situation, give case advice or promise an outcome. Requests set store:false and have no research tools. OpenAI does not use API content for model training by default unless the account opts in. This setting is not a promise of zero retention: default abuse-monitoring logs can retain content for up to 30 days. This chat’s requests disable prompt-cache creation and reuse; where caching is used, it can retain encrypted application state for up to 24 hours. Disabling caching does not eliminate abuse-monitoring or legally required retention. Separately approved OpenAI Zero Data Retention controls have limitations, including safety or legal exceptions. Free-text processing is subject to browser-verification and usage controls; those controls do not establish zero provider retention. The FAQ does not switch to another provider.

Usage and security information

A signed first-party browser cookie identifies this anonymous preview for its question and budget limits. It is Secure, HttpOnly and SameSite Strict, can last up to one year, and contains no conversation. Persistent pseudonymous counters enforce the limit across refreshes; short-lived IP counters discourage abuse. Clearing cookies or using private browsing can create a new browser identity, so this is not a claim to identify a unique person. Daily aggregate budget and alert records contain no conversation text.

Browser verification

Before a paid AI request, Cloudflare Turnstile may check the browser and ask for an interaction. Its browser checks process technical information under Cloudflare’s privacy terms. The FAQ does not send your question or case summary to Cloudflare; its verification request contains the challenge token and service secret. Challenge success does not prove a unique human. If verification or budget controls fail, no model request is made and the chat can show an approved AIA overview instead.

Hosting and AI services may keep operational metadata, such as request times, status and token usage, under their policies. The FAQ must not use analytics or logs that capture conversation content. A budget-limit event can be recorded for operator notification without question text or visitor identifiers.

Cloudflare Turnstile privacy information

The AIA app is separate

The app holds case records for ongoing work. This temporary FAQ’s handling does not describe app case retention. Review the app’s privacy terms before adding personal information there.

Questions about privacy can be sent to alex@aiadvocate.ca.

OpenAI API data controls and retention