AI governance policy

The one-line version

AI helped build this website and write its words. AI never touches the names you save.

Where AI is used

• Writing code. The HTML, CSS and JavaScript on this site were drafted with Claude, an AI assistant made by Anthropic, and reviewed by a person before publication.
• Writing content. Page copy, including the white paper and the first drafts of the legal pages, was produced with the same assistance and edited by a person.
• Design. Layout, colour and typography choices were developed with AI assistance.

Where AI is not used

• No name, relationship, birthday or note you enter is sent to an AI system. The app makes no AI request at all.
• Nothing you enter is used to train any model, ours or anyone else's.
• No AI ranks, scores, profiles or infers anything about you or the people in your book.
• No automated decision is made about you. There is nothing to decide, because we hold no account.

The commitments we hold ourselves to

1. Disclosure. Every page of this site carries a visible notice that AI was used to build it. We do not present AI-assisted writing as though a person wrote every word unaided.
2. Human review. No AI-generated code or copy reaches this site without a person reading it first. A person is accountable for everything published here, and AI assistance is not an excuse for an error.
3. Data minimisation. If an AI feature is ever added to the product, it must work on the device or not ship. We will not build a feature that requires uploading your book to a model.
4. Consent before change. Any future AI feature that would process your data will be off by default, explained in plain language, and switched on only by you.
5. No training on user data. Your data will never be used as training material, and we will not sell or supply it to anyone who would use it that way.
6. Accuracy. AI-written text can be confidently wrong. We check factual claims, and the legal pages get reviewed by a qualified human before they are relied on.
7. Bias and fairness. The product sorts people into circles you define yourself. We will not add AI that categorises, ranks or judges the people in your book.
8. Accountability. Project Research LLC is responsible for this product regardless of which tools were used to build it. Questions and complaints go through the contact page and are answered by a person.

Third-party AI

Two Google services can load on this site if you accept on the consent banner: Google Analytics and Google AdSense. Google may apply its own machine learning to the traffic and advertising data it collects. That data is limited to page views and ad interactions. It never includes the contents of your book, which never leaves your device to begin with.

Standards we are working toward

This policy is written with the NIST AI Risk Management Framework and the transparency duties of the EU AI Act in mind. Jackdaw Book is a low-risk consumer product with no automated decision-making, so most obligations in those frameworks do not attach to it. We publish this anyway, because a product that asks for trust should say what it does.

Changes

Material changes to this policy will be announced on this page before any new AI feature ships.