--- title: Privacy Policy slug: privacy description: How Arbory handles what you add, what it sends to the companies that help run it, and the controls you have. updated_at: '2026-11-15' --- Arbory ("we") is run by Michael Kemp in Des Moines, Iowa. It tracks what you read, listen to and watch. We store what you add and the notes you write so the app works for you, and we count things across readers. We don't read your notes or chats, sell your data, use it for advertising, or use it to train AI models. Arbory is for people 18 and older. ## What we collect - **From Google sign-in:** your name, email address and profile photo. We don't get access to your Gmail, Drive, contacts or anything else in your Google account. - **From email sign-up:** your email address, the name you enter, and a password. Supabase stores the password only as a scrambled hash; we never store it. Arbory sends no sign-up emails. - **What you add:** books, publications, documents, podcasts and videos, sessions, shelves, notes, quotes, photos you upload, your chats with the assistant, and your settings. - **If you connect an AI assistant:** which assistants you've connected, when you connected each one, and the last day each used Arbory. We also record which changes came from an assistant. - **Usage data:** pages you visit in Arbory, how you arrived (referrer and campaign tags), your browser type, and which features you use. When you're signed in, this is linked to your account. - **Feedback and the contact form:** what you send, and your name and email for the contact form. ## How we use it We use your data to run Arbory for you, to help when you ask for support, and to see which pages and features get used so we know what to fix and build. We don't sell your data, use it for advertising, or use it to train AI models. ## Who can see your notes and chats Nobody at Arbory reads your notes or chat messages. Our support tools show how the assistant behaved, not what you wrote. If you choose to share a chat with support, we can see that one chat for 7 days, and every view is logged. There is no admin override. ## Counts across readers We count things across readers, like how many readers finished a title. A count is only kept when at least 5 readers are behind it, and it never says who. ## Service providers These companies receive your data so Arbory can run. Each gets only what it needs, and none may use it for anything else. - **Supabase (sign-in and database):** creates your account and stores your data in the United States. It keeps daily backups for 7 days. - **Replit (hosting and file storage):** runs Arbory and stores photos you upload, in the United States. It keeps our server logs for 30 days. - **Anthropic and OpenAI (AI models):** power chat, note formatting, summaries, photo-to-quote and podcast guest lists, and build the search index for your notes. - For chat and these tasks, your messages, the relevant parts of your library and any photo you use for a quote are sent to produce a reply. - For the search index, the text of your notes and quotes, your comments on quotes and what you search for are sent, never who you are. When you delete a note, its entry in the index is deleted with it. - Anthropic deletes what we send within 30 days. If its safety systems flag a request as misuse, it may keep that request for up to 2 years. OpenAI keeps what we send for up to 30 days to check for abuse, then deletes it. Either may keep data longer if the law requires it. Neither uses it to train models. - **Plausible (analytics):** counts page visits without cookies, on servers in the EU. It sees the page (with the IDs of your works and chats removed), the referring site, your browser type and approximate location. It never stores your IP address. Arbory also looks things up in public sources, described below. Those lookups send a title, name, link or identifier, never who you are. - **Book and paper lookups:** fill in details about books and papers from public book and paper databases. We send a title, author, ISBN or DOI. - **Podcast and video lookups:** fill in details about episodes and videos you add from a link. Each link is sent only to the service it came from, and we send only the link or the ID in it. For an episode, we look up its show in Arbory's own list of public podcasts, built from a public podcast directory, and read the show's public feed from wherever it's hosted, to fill in the date, length and episode number. If the show isn't on that list, we may ask a public podcast directory about it. You can also paste a show's feed link and pick an episode. Members-only and private feeds are never fetched. - **Episode search by name:** if you name a podcast episode in chat or through a connected assistant without a link, our server first checks Arbory's own catalog and list of public podcasts, and may search a public podcast directory if those can't answer. The same goes for an episode link from a podcast app that gives us only the episode's title: to find its show, we look that title up the same way (with the show, if you typed one) and let you pick from the closest matches. We send the title, plus the show if you named one. This never happens for anything you mark private, or for books, publications or documents. - **Daily feed checks:** once a day, Arbory checks the public feeds of shows in its catalog for new episodes. No reader's account is part of that request. - **Guest lists:** we use the names a show's feed gives, when it gives them. Otherwise an AI model reads the episode's title and public notes from the show's feed. Nothing about you is sent, and we keep only the names. A guest list may be wrong. You can fix it on your copy, which doesn't change anyone else's. - **Show and publication names:** come from public sources like podcast directories, a show's feed, and public book, paper and reference databases. - **YouTube:** Arbory uses YouTube API Services to confirm that a video you add is public and to fill in its title, channel and thumbnail. We send YouTube only the video's ID, never who you are. We keep that information for at most 30 days at a time and check it with YouTube again before then. If a video stops being public, we delete what YouTube gave us. YouTube's handling of data is covered by the [Google Privacy Policy](https://policies.google.com/privacy). ## AI assistants you connect You can connect Arbory to an AI assistant like Claude or ChatGPT. Before you allow it, Arbory shows you what it can read and change. A connected assistant can read your library (titles, shelves and where you left off), your sessions, your notes and quotes in full, and the links you've saved, private ones included. It can also look up your stats and search your notes and quotes. If the assistant asks for them, it also gets your email address and name, and Arbory tells you before you allow it. It can add, log, edit, move and delete works, sessions and notes. It can't read your chats with Arbory's assistant, change your account settings, or delete your account. What it reads leaves Arbory and goes to the assistant's provider, under their privacy policy, not ours. Arbory gets only the requests the assistant sends, never your conversation with it. It stays connected until you disconnect it on the Connect page, which shows when each one was connected and last used. Disconnecting ends its access right away. When an assistant uses Arbory, we log which tool it used, whether it worked and how long it took, never what it sent or got back. ## Google user data Arbory asks Google only for your name, email address and profile photo. We use them only to create and secure your account and to show your name in the app. We don't share them with anyone except the service providers that run sign-in and storage, and we never use them for advertising. Arbory's use of information received from Google APIs adheres to the [Google API Services User Data Policy](https://developers.google.com/terms/api-services-user-data-policy), including the Limited Use requirements. ## Links you add When you add a public link, including one you attach later to something you entered by hand, our server opens the page to fill in its title and details. YouTube videos are the exception: we ask YouTube about them instead, as described below. The request doesn't carry your account, cookies or anything else that says who you are. Our server identifies itself as Arbory-Fetcher; [arboryapp.com/fetcher](/fetcher) explains what it fetches. If that link leads to something already in Arbory's shared catalog, your entry moves to the shared title, with your sessions and notes. Two links to the same thing can end up as one shared title. An episode joins the shared catalog once its show's own feed lists it and the show is in a public podcast directory. Episodes from a feed you paste that isn't in one stay visible only to readers who added them. Other readers can find a shared title, but never who has it. If you add a private link, like a Google Doc or Notion page, or tick "Treat this link as private", we store it as a link only. We never open it or look it up, and we leave it out of everything Arbory sends to its AI providers. It's shown only to you: in the app, in your export, and to an AI assistant you connect, which then sends it to that assistant's provider. A link you type into a chat message or a note is part of that text, and goes wherever the text goes, including to our AI providers. A video link is checked with YouTube to confirm the video is public. To check, Arbory sends YouTube only the video's ID, nothing about you. If YouTube says the video is private, unlisted or missing, it's saved just for you and we keep nothing from YouTube. If we couldn't get an answer from YouTube, it's saved just for you and we check again a few times over about a day, then stop. We only check again when we couldn't get an answer: links you mark private, and videos YouTube says are private, unlisted or missing, are never sent again. If it turns out to be public, your entry moves to the shared title, with your sessions and notes. Playlists, channels and other YouTube links are saved as private without being checked. ## Cover images Some covers, like video thumbnails, podcast artwork, and images from articles and book databases, load in your browser straight from the site that hosts them. That site can see your IP address, that the image was loaded from Arbory, and which image it was. It doesn't learn who you are in Arbory. ## Cookies and storage Arbory doesn't use advertising or tracking cookies. Your browser stores data to keep you signed in and to make the app load quickly. ## Legal requests We disclose your data only when we receive valid legal process, like a subpoena or court order, and we tell you first unless the law forbids it. Legal requests go to [support@arboryapp.com](mailto:support@arboryapp.com). ## If something goes wrong If a security breach affects your data, we'll tell you directly, without undue delay, along with what happened and what we're doing about it. ## Your controls - **Export:** Settings → Your data → Export downloads your library, shelves, sessions, notes, chats and settings as a JSON file. It doesn't include Recently deleted. - **Fix:** you can edit or delete the works, sessions, notes and quotes you add. Some details of a shared title can be set on your copy without changing anyone else's: its cover, length and guests, and its show if it doesn't have one. For anything you can't change yourself, like your name or email, write to us. - **Recently deleted:** when you delete a work, session or note, in the app or through a connected assistant, it goes to Settings → Recently deleted for 30 days, where you can restore it. After that it's erased within a day. - **Edits by an assistant:** when a connected assistant edits a note, or the details of a work you added, we keep the earlier text for 30 days, then erase it within a day after that. - **Connected assistants:** see and disconnect them on the Connect page. - **Delete:** Settings → Your data → Delete account permanently erases your library, sessions, notes, chats, feedback, uploaded photos, everything in Recently deleted, your connected assistants and your sign-in account, right away. Two things stay, with nothing that links them to you: records of pages visited and sign-up steps, with your account removed from them, and counts across 5 or more readers, which drop you the next time they're rebuilt, within a day. Messages you sent through the contact form aren't part of your account; write to us and we'll delete them. - **Backups:** our database provider keeps daily backups for 7 days. Anything erased leaves those backups within 8 days. - **Copies at AI providers:** what we sent to an AI provider isn't erased when you delete something in Arbory. Each provider deletes it on its own schedule: within 30 days at Anthropic and 30 days at OpenAI, or up to 2 years at Anthropic if a request was flagged as misuse. - **Questions or requests:** [support@arboryapp.com](mailto:support@arboryapp.com) ## Age Arbory is for people 18 and older. If we learn an account belongs to someone younger, we'll delete it. ## Who we are Arbory is run by Michael Kemp, Des Moines, Iowa. Questions: [support@arboryapp.com](mailto:support@arboryapp.com). ## Changes We'll update this page and its date. If a change matters, we'll tell you in the app.