What we hold, and what we never touch.

MapLit (“we”) is an AI literature mapping tool. This policy covers the MapLit web application and the accounts, libraries and files it holds. It applies whether you use MapLit signed in or anonymously.

MapLit is operated by an independent founder rather than an established company. The formal operating entity, its registered address, and the supervisory authority you may complain to will be named here before public launch.

Account

Your email address. Nothing else. There is no password, no name field, no avatar, no phone number. Sign-in is a one-time link sent to your inbox.

Your library

The content you create and import, so it can sync across your devices:

• Library and branch names, the map structure, and node positions.
• Paper metadata — title, authors, year, abstract, DOI or arXiv id, venue and page numbers.
• AI-generated analysis of each paper (the “concept card”) and of the library as a whole.
• Your reading work — highlights and the passages they quote, per-paper notes, your conversations with the AI assistant, and which page of a paper you last read.

Files

The PDFs you upload, stored under a key derived from your account id and the file’s content hash. We do not store the original filename with the file, though the filename is sent once to our AI provider during processing (see section 4).

Usage records

One row per AI request: which feature was used, the model, token counts, cost, and the time. This is how we keep AI spending under control and show you your own usage. The content of your prompts and the AI’s replies is not stored in these records.

If you use MapLit without an account, the trial limit is enforced per IP address, so your IP address is recorded in that usage row instead of an account id. Signed-in usage rows contain no IP address.

• No analytics, and no behavioural tracking. We do not profile you, follow you across sites, or measure what you click. The one third-party tool that runs in your browser is crash reporting (see section 6) — and it is scrubbed of your cookies, your content and your session, so it only ever sees a technical error.
• No advertising, and no profiling or automated decisions with legal effects.
• We do not sell, rent or share your data. The only third parties involved are the infrastructure providers listed in section 6, acting on our instructions.
• No fonts phoning home. Typefaces are bundled with the app at build time, so your browser makes no request to a font provider.

MapLit’s core function — reading a paper and placing it on a map — requires sending text to OpenAI, our AI provider. Depending on the feature, what is sent includes:

• Paper text extracted from your PDF in your browser (for metadata and structure extraction, and for AI highlights), together with the file’s name. The PDF file itself is never sent to OpenAI.
• Paper title, authors, year and abstract.
• Your existing library structure — branch names and neighboring papers — so a new paper can be placed sensibly.
• Your messages to the AI assistant, the conversation history, and, while reading, the text of the page you are on plus your recent highlights.

Your email address and account id are never sent to OpenAI. Requests are made from our server, not your browser.

Under OpenAI’s API data-usage policy, content submitted through the API is not used to train their models by default, and is retained for a limited period for abuse monitoring. That is OpenAI’s commitment rather than ours; we will confirm the exact terms and any data-processing agreement in force before public launch, and state them here.

MapLit runs on global infrastructure. Depending on the component, your data is stored or processed in Singapore, Japan, the United States, and other locations operated by the providers in section 6. If you are in China, the European Economic Area, the United Kingdom, or anywhere outside those regions, using MapLit means your data is transferred abroad.

Because of this, we ask for your explicit agreement to overseas storage and processing when you sign in — separately and unticked, not buried in a bundle. If you do not agree, do not sign in; you can still use MapLit anonymously in your browser, in which case your library stays on your device until you choose to sign in.

The formal transfer mechanism for users in the EEA/UK (standard contractual clauses) and the filing route required under China’s PIPL for cross-border transfers are being put in place with counsel before public launch. We would rather state that plainly than imply a compliance posture we have not yet completed.

We also fetch public metadata from arXiv and Crossref when you add a paper by link or DOI. These requests carry no account identifier.

One honest detail: when MapLit opens an arXiv PDF, your browser normally fetches it from arxiv.org directly, so arXiv sees your IP address, exactly as it would if you visited arxiv.org yourself. We fall back to fetching through our own server only when the direct request is blocked.

We do not use Stripe, PostHog, or any analytics provider today. If that changes, this list changes first.

Cookies

MapLit sets only essential cookies — the Supabase authentication cookies that keep you signed in (sb-<project>-auth-token and a short-lived sign-in verifier). They last up to 400 days and are removed when you sign out. There are no analytics or advertising cookies, which is why MapLit does not show you a cookie banner: there is nothing to consent to beyond keeping you logged in.

On-device storage

MapLit is local-first, so your browser also keeps a working copy of your library — the map and papers in local storage, and your highlights, notes, conversations and PDF files in IndexedDB. This is what lets the app stay fast and keep working offline.

On a shared computer: signing out erases that local copy, and so does signing in as a different person. Your library is never inherited by the next user of the browser.

We keep your data for as long as your account exists. Items you delete inside the app (a paper, a branch) are marked deleted so that deletion can sync safely across your devices, and are removed with the account.

Deleting your account — from your account page — erases your PDF files from our file storage, then deletes your account and every row belonging to it: profile, libraries, papers, map, highlights, notes, conversations, reading positions, usage records and consent record. It is immediate and cannot be undone, so export first if you want to keep anything.

One exception, stated plainly: usage rows created while you were not signed in are keyed to an IP address rather than to your account, so account deletion cannot find them. They contain no content — only a feature name, token counts and cost — and they are deleted automatically 30 days after they are written.

Depending on where you live, you have the right to access, correct, export, and erase your personal data, to restrict or object to its processing, and to withdraw consent. In practice, MapLit builds most of these into the product rather than into an email queue:

• Access and portability — export your whole library at any time as BibTeX, CSL-JSON, RIS, or a complete JSON file from your account page.
• Erasure — delete your account from the same page.
• Withdrawing consent — deleting your account withdraws your consent to overseas processing and removes the data it covered.

For anything else — correction, restriction, objection, or a complaint — write to privacy@maplit.ai. You also have the right to complain to your local data protection authority.

• Every table enforces database-level row security: a query can only ever return rows owned by the account that made it.
• PDF files are stored under a per-account prefix and can only be downloaded through a short-lived signed link, issued after we verify that the file is yours.
• Uploaded files are re-hashed on our server and rejected if the contents do not match, and re-verified again when your browser downloads them.
• Everything travels over encrypted connections.

No system is perfectly secure. If you believe you have found a vulnerability, please write to privacy@maplit.ai before disclosing it publicly.

MapLit is built for researchers and is not directed at children. Do not use MapLit if you are under 16 (or the minimum age of consent in your country, if higher).

If we change how we handle your data — a new provider, a new purpose, a new category of data — we will update this page, change its version, and ask for fresh agreement where the law requires it. The date at the top always reflects the version in force.

privacy@maplit.ai