# AI policy

URL: https://inboxreplyai.com/ai-policy/

What InboxReplyAI may say (only from your approved sources), what it will not say, how replies are disclosed, when it hands off, and how mistakes get fixed.

This page says how the InboxReplyAI assistant is meant to behave. It describes the design we demo and are building.

## What the AI may say

It answers only from the sources you approve: your website, documents, price list and FAQs. If the answer is in those sources, it can give it. If not, it does not guess.

## What it will not do

- **No medical, legal or financial advice.** It hands those chats to your team.
- **No prices that are not in your sources.** It does not invent or estimate a price.
- **No promises or guarantees** about results, outcomes, timing or availability that your sources do not state.
- **No pretending to be a person.**

When it cannot answer, it says it will get a person, instead of guessing.

## Disclosure

We recommend telling customers they are talking to an automated assistant. For example: "You are interacting with an automated assistant. A person can step in at any time."

Some laws require disclosure in some situations. A growing number of US states have chatbot laws, with different triggers. For example, California's BOT Act addresses using a bot online to mislead a person about its artificial identity in a commercial transaction, and a clear disclosure is a safe harbor. Maine has a 2025 chatbot disclosure law. Meta's rules also say automated chats must disclose where the law requires it.

This is a summary, not legal advice. You are responsible for the rules that apply to your business.

This is a plain-English policy for an early-stage product; we will update it as the product launches.

## Handoff rules

The assistant hands the chat to your team when:

- It is not sure of the answer.
- The customer asks for a person.
- The topic is medical, legal, financial or a complaint.
- The question needs something outside your approved sources, such as a custom quote.

Your team sees the full conversation. How handoff is set up is covered in [How it works](/how-it-works/).

## Instagram rules we follow

- It replies only to people who messaged your business first.
- It replies only within Instagram's messaging window (24 hours for most user actions).
- No cold DMs and no broadcasts.

We are not affiliated with Instagram or Meta.

## When the AI gets it wrong

Tell us, or tell your team, and we fix the source or the rule that caused it. You can edit or remove any source at any time. We look at reported mistakes ourselves and change the setup. Send examples to [hello@inboxreplyai.com](mailto:hello@inboxreplyai.com).

## Related

[Security](/security/) · [Editorial policy](/editorial-policy/) · [About](/about/) · [Book a demo](/demo/)


Book a demo: https://inboxreplyai.com/demo/
