AI Features Notice

This notice explains Hunt's AI features, Summarize and Ask AI in the sidebar: what is sent, where it goes, how long it is kept and what the AI can and cannot do. It adds to our Terms of Service and Privacy Policy.

What the AI features do

Summarize makes a short summary of the page you are on, or of a video from its captions.
Ask AI answers your questions in a chat beside the page. With "Use this page" on, it can answer about the page you are on.

You need a Hunt account to use them. There are daily limits on the number of summaries and questions, shown in the sidebar.

You are always talking to an AI

Summaries and answers are made by a computer model, not by a person. They are shown in the AI sidebar as AI output, and are not checked by anyone at Hunt before you see them.

When a page is read

Hunt only reads a page when you ask it to. It reads the page when you press Summarize, or when you send a question with "Use this page" switched on. Before reading a page for Ask AI, Hunt asks you to allow that site. Hunt does not read pages in the background for AI.

What is sent, and where

When you use an AI feature, Hunt sends:

the page's title, its web address and its text, or the video's captions (when you asked about the page);
your question and up to the last 10 messages in the chat;
your account's session, so we can apply daily limits.

This goes to our AI service, which runs on Cloudflare Workers AI. The model is currently Meta's Llama 3.1 (8B Instruct), run by Cloudflare for us. We may change the model; this notice will name the current one.

How long it is kept

We do not store your questions or our answers after the answer is made.
We keep a cache of summaries for 7 days, so the same content is not processed twice. The cache is keyed by the content, not by you.
The chat is kept only in the open sidebar on your device. New chat clears it.
We count how many requests you make each day, to apply the daily limits.

No training on your data

We do not use your pages, questions or answers to train AI models. Cloudflare's terms for Workers AI say it does not use the data we send to train models either.

Please be careful with sensitive pages

Don't use AI features on pages with information you would not want to share, such as banking, medical records or private messages. If you do, that text is sent to our AI service like any other page.

Limits of AI

AI can be wrong. It can misread a page, leave things out, invent facts, or give answers that are out of date.
AI output is not advice. Don't rely on it for medical, legal, financial, tax or safety decisions. Check the original page or ask a qualified person.
A summary is not a substitute for the article. It may miss important details. Publishers' content belongs to them; read the original for the full story.
The model has its own safety training, but it may not always avoid harmful output. If you see harmful output, report it to hello@huntbrowser.com.

Your choices

AI features only run when you start them. You can simply not use them. You can clear the chat with New chat at any time. To ask about your data, write to hello@huntbrowser.com.

Rules for using AI features

Don't use AI features to create content that breaks our Terms of Service, to get around a website's paywall or rules, or to make content that pretends to be a real person or event. Don't send requests automatically or at scale.