# Content licensing and permitted use

What you may do with the scholarly content this API returns.

## Rights by content type

| Content | Store | Cache | Show end users | Redistribute | Train models |
|---|---|---|---|---|---|
| Paper metadata (title, authors, DOI, venue, year) | yes | yes | yes | yes, with attribution and DOI link | no |
| Abstract | yes | 30 days | yes, with attribution | no | no |
| `tldr` machine summary | yes | yes | yes, labeled "Summary by SciSpace" | no | no |
| Open-access full text | per its licence | per its licence | per its licence, licence notice preserved | per its licence | no |
| Licensed (non-OA) full text | no — processing only | 24 hours | snippets ≤ 200 words with citation | no | no |
| Generated answers and citations | yes | yes | yes | no, not as a dataset or a competing corpus | no |
| Topics and extraction results | yes | yes | yes | no | no |
| Your own uploaded documents | yours | yours | yours | yours | yours |

## Attribution

Public-facing surfaces that display paper content must show the paper title, the first author, the
venue, and a resolvable DOI link. Open-access content must carry its licence label (for example
CC BY 4.0). Products built on generated answers must carry "Powered by SciSpace" or equivalent.

## Bulk and derivative use

Systematic downloading to build a competing corpus or search index is not permitted. Bulk metadata
use for research is — talk to `sales@scispace.com` first so we can raise your limits rather than
throttle you.

## Data sources

The corpus aggregates metadata from Crossref, PubMed/PMC, arXiv, and OpenAlex, plus direct publisher
agreements. Rights vary per record: check `open_access` on each Paper rather than assuming a
blanket licence.

## What is explicitly allowed

- Building a product that searches, summarizes, and cites papers for your users
- Storing metadata and generated answers in your own database indefinitely
- Showing quoted evidence with a citation and a link to the source
- Exporting bibliographies for your users' own reference managers

## What is explicitly not allowed

- Rehosting full text, or mirroring the corpus
- Selling generated answers or extracted rows as a dataset
- Training or fine-tuning a model on returned content
- Stripping attribution, DOI links, or licence notices

## Changes to this policy

Material changes are announced 30 days ahead by email and in the changelog.

## Related

papers · data-privacy · migrate-from-openalex · batch-extraction