scime · Latentic Labs
Terms of service
The rules for using the scime atlas.
Effective October 21, 2026
1. The service
The scime atlas (atlas.scime.ai) is provided by Latentic Labs. It shows accepted papers from major machine learning conferences on a map, finds published work related to an idea you describe, and offers optional AI summaries. It is an early prototype provided free of charge, and features may change or stop.
2. Accounts
You can use the atlas without an account. Signing in is optional and uses your scime.ai account, which is covered by scime.ai's terms. Without an account you can browse every venue on the map and in Flow and place your ideas with their nearest papers, up to a daily number of placements. Signed-in users can also save ideas, switch on weekly email alerts, create share links, see summaries of what the closest papers found, and get a larger daily AI allowance. The email list for the scime Dispatch needs no account; every email has a one-click unsubscribe link. Reading lists are kept only in your browser; clearing your browser data removes them, and we cannot restore them. The service is not directed to children under 14.
3. Acceptable use
- Do not try to break, overload, or get around the limits of the service, including the daily AI limit.
- Do not scrape the service in bulk or use it to build a competing dataset.
- Do not submit material you are not allowed to share, such as manuscripts under confidential review. Submitting an idea to an AI feature sends it to OpenAI.
- Do not use the service to break the law or the rules of your institution or a conference.
We may limit or suspend access to deal with misuse, security problems, or legal obligations.
4. Your content
You keep all rights to the ideas and feedback you submit. When you create a share link, anyone who has the link can see what it shows until it expires or you delete it; only share what you are allowed to share. You allow us to process them only as needed to run the features you use, as described in the privacy policy. We may use feedback to improve the service. We collect anonymous usage statistics to understand how the service is used; the privacy policy explains what they contain and how to opt out.
5. Paper data and AI output
Paper titles and links come from public accepted-paper lists. Recent preprints come from the arXiv API (thank you to arXiv for use of its open access interoperability); preprints are not peer reviewed. Each paper belongs to its authors and publisher; follow the link to read the original. The collection is incomplete: some years and venues are missing, and grouping into areas and topics is automatic and can be wrong.
AI output is generated from short abstract excerpts and can be wrong, incomplete, or miss important prior work. "Low overlap" does not mean an idea is new. Check the original papers, and follow your institution's and venues' rules on using AI. We do not guarantee novelty, acceptance, or any research outcome.
6. Disclaimer and liability
To the extent permitted by law, the service is provided "as is" without warranties of availability, accuracy, or fitness for a particular purpose, and Latentic Labs is not liable for indirect or consequential losses arising from its use. Nothing in these terms limits rights or liability that cannot be limited by law, including mandatory consumer protections in your country.
7. Changes and contact
We may update these terms when the service changes and will post the new version here with its effective date. Questions: contact@latenticlabs.com. These terms are governed by the laws of the Republic of Korea, without depriving you of protections under the mandatory laws of your country of residence.
