scime · Latentic Labs
Privacy policy
What the scime field atlas collects, why, and where it goes. Short version: no accounts, no cookies, anonymous usage counts, and your idea text is used on our own server to find related papers without being stored. It goes to a third party only if you choose an AI check.
Effective October 1, 2026 (updated for on-server semantic search)
1. Who we are
The scime field atlas (atlas.scime.ai) is operated by Latentic Labs, Republic of Korea ("we"). It is a map of accepted papers from major machine learning conferences, where you can place a research idea and see related published work. There are no accounts: you never sign in, and we do not ask for your name or email.
2. What we collect
Anonymous usage statistics. To understand how many people use the atlas and which parts help, the page reports what you do, for example "opened the Flow view", "placed an idea, which landed in Large language models › Reasoning", "saved an idea", or "opened a paper link". With each report we record:
- the action and a few details about it (such as the area and topic an idea landed in and how many words it had, or which venue filter was chosen);
- a random visit number created when the page loads. It is kept only in the open page and disappears when you close or reload it;
- a daily-rotating visitor code: a one-way hash of your IP address and browser identification combined with a random value that we replace every day (UTC) and then delete. It lets us count unique visitors per day, but it cannot be turned back into your IP address and cannot link your visits across days;
- on the first page view only: the website you came from (its domain name only, such as reddit.com), the
utm_sourcetag in the link if there is one, and your device type, browser family, and a rough screen size (small, medium, large).
We do not record the text of your ideas in these statistics, your IP address, or anything that identifies you.
Your ideas. When you place an idea, its text is sent to our server, which turns it into numbers with an open-source language model running on that server and compares it with the papers. The text is used only for that search: it is not written to our database or logs, and it is not sent to anyone else. Saved ideas stay in your browser; to show how many recent papers are near them, the page sends their text for the same kind of search. Your idea text is stored by us or sent to a third party only in these cases:
- you press Check novelty with AI: the idea is sent through our server to OpenAI (section 4). Our server does not keep it;
- you rate a result as Not relevant or Wrong cluster: the idea is stored with the rating so we can check the result;
- you send written feedback and tick Attach the idea I placed.
Feedback you write is stored with the time and the random visit number of that page load.
AI usage. For each AI request we record its type, the model, the number of tokens, success or failure, and the time. To enforce daily limits we keep a counter per daily-rotating visitor code. We do not keep AI prompts or answers.
Your IP address is used briefly in server memory to prevent abuse (rate limits) and to compute the daily visitor code. It is not written to our database.
3. Why and on what basis
| Purpose | Data | Legal basis |
|---|---|---|
| Understand and improve how the atlas is used | Anonymous usage statistics | Legitimate interest in running and improving the service. The data is designed not to identify you, and you can opt out (section 6). |
| Find related papers | Idea text, processed on our server by an open-source model and not stored | Providing the feature you request |
| AI novelty check and topic guide | Idea text you submit, paper abstracts, request records | Providing the feature you request; you start each request yourself |
| Fair use limits and abuse prevention | Daily AI counters, IP address in memory | Legitimate interest in keeping the service available and affordable |
| Check result quality and improve the service | Feedback and ratings, with the idea text when you choose to include it | Legitimate interest; sending feedback is optional |
We do not sell data, show advertising, use third-party analytics or tracking, or build profiles of individuals.
4. Who receives it, including transfers outside Korea
We use the following providers to run the service. They process data on our behalf or as needed for their function, under their own security and data-protection terms.
| Recipient | Country | Data | Purpose | When and how | Retention |
|---|---|---|---|---|---|
| Amazon Web Services, Inc. | United States | All data described in section 2 | Server hosting and storage | Continuously, over encrypted connections | As in section 5 |
| OpenAI, L.L.C. | United States | The idea text you submit and excerpts of related paper abstracts | Generating the AI novelty check and topic guide | Only when you press an AI button, over an encrypted connection | Under OpenAI's API data policies (openai.com/policies). OpenAI does not use API inputs to train its models by default. |
Our server itself runs on Amazon Web Services in the United States, so the search step also processes your idea text there (without storing it). If you do not want your idea sent to OpenAI, do not use the AI buttons; everything else works without them.
We disclose information to authorities only when the law requires it.
5. How long we keep it
- Idea text sent for searching: not stored; discarded as soon as the search is done.
- Anonymous usage statistics: 13 months, then deleted automatically.
- The random value behind the daily visitor code: deleted after its day ends.
- Feedback and ratings: up to 3 years, then deleted automatically.
- AI request records: 1 year. Daily AI counters: 90 days.
- Server error logs: rotated automatically and kept for a short time.
- Backups, if made, are overwritten or deleted within 30 days.
6. Your choices and rights
- Opt out of usage statistics: turn on "Do Not Track" or Global Privacy Control in your browser. The atlas then records no usage statistics for you.
- Ideas and reading lists live only in your browser. Clear them by clearing this site's data in your browser settings.
- Feedback: because we do not know who you are, we usually cannot find your feedback by ourselves. Email us with the text and roughly when you sent it, and we will delete it.
Depending on where you live, you may have rights to access, correct, delete, or restrict processing of personal information, and to object to processing based on legitimate interest. Email us (section 10); we respond within 10 days. If you are in the EU, the UK, or another region with its own data-protection law, you can also complain to your local data-protection authority. In Korea you can contact the Personal Information Protection Commission (pipc.go.kr) or the Personal Information Infringement Report Center (privacy.kisa.or.kr, phone 118).
7. Cookies and browser storage
The atlas sets no cookies for visitors. Your browser stores, on your device only, the ideas and reading list you save and cached AI topic summaries, so they are there next time. We never read this storage for statistics. (The site operator's own dashboard uses one sign-in cookie that visitors never receive.)
8. Security
Connections are encrypted (HTTPS). The database is not reachable from the internet, and access to the server is limited to the operator. Usage statistics are designed so that even we cannot tell who you are.
9. Children
The service is meant for researchers and is not directed to children under 14. If you believe a child has sent us personal information in feedback, email us and we will delete it.
10. Contact and privacy officer
Privacy officer: Representative, Latentic Labs
Email: contact@scime.ai
Please do not include passwords or confidential manuscripts in your message.
11. Changes
If we change how we handle personal information, we will update this page and its effective date, and announce material changes on the site before they take effect.