Uses the articles whose publisher recorded your ORCID iD with Crossref. Missing some? Paste their DOIs.
What a match means
A match means the article's DOI is on one of these lists. Sci-Hub and LibGen's scimag collection are pirated copies of journal articles; LibGen's database records the article's DOI, title and file checksum, and when it was added. A match does not show that any company downloaded the article or trained a model on it. Not every DOI on Sci-Hub's 2017 list resolves, and a DOI missing from these lists may still be in a later or different copy.
Who owns the copyright. Many journal authors assigned or exclusively licensed their copyright to the publisher when the article was accepted. If you did, the publisher, not you, may be the owner entitled to sue; open-access articles under a Creative Commons license are a different case again. Check your publishing agreement.
Registrations. Statutory damages need a U.S. registration made before the copying began, or within three months of publication (17 U.S.C. § 412). For articles that usually means the publisher's registration of the journal issue, which covers the articles in it that the publisher owned; some authors register their own articles. The registration report checks both against the Copyright Office's serial registrations (1978–2025) and TX registrations (to Dec. 2025), and compares the dates with the article-copying events below. Matches are candidates to confirm against the linked Office record: an issue registration is matched by journal title and issue date, and whether it covers a given article depends on who owned that article when the issue was registered.
AI companies and these collections
- Company filingMeta admits obtaining data from LibGen, Anna's Archive and “SciHub” to develop and train some Llama models. Elsevier v. Meta, No. 1:26-cv-03689 (S.D.N.Y.), Answer ¶82 (Dkt. 78, July 13, 2026).
- AllegationMeta: a Meta researcher torrented all of LibGen's SciMag collection, about 81 million articles. Hobbs v. Meta, No. 1:26-cv-04314 (S.D.N.Y.), complaint ¶76 (May 22, 2026); see also Elsevier v. Meta, complaint ¶¶82, 88.
- AllegationNVIDIA downloaded from LibGen, Sci-Hub and Z-Library, on information and belief. Nazemian v. NVIDIA, No. 4:24-cv-01454 (N.D. Cal.), first consolidated amended complaint ¶58 (ECF 235, Jan. 16, 2026).
- AllegationAnthropic: a plaintiffs' expert in Bartz v. Anthropic states that torrent code Anthropic produced lists “scimag_torrents” alongside the LibGen torrents (Dkt. 560-6 ¶25, Jan. 21, 2026). No court has ruled on it.
This page is general information, not legal advice. Article lists come from Crossref's open metadata; we do not call the ORCID API. Searches are logged without IP addresses, as described in the privacy policy. Sources and checksums: Methodology.