SCREEN Systematic Claim Recomputation & Evidence Evaluation Network
SCREEN is a proposed open infrastructure that automatically checks every article published in referenced mathematics journals (AMS, EMS, Springer, Taylor & Francis, etc.) as soon as it appears, together with the large corpora of results now produced by AI systems. SCREEN does not issue opinions on proofs and does not grade articles. AI models redo what can be redone: numerical values, small cases of general statements, citations, the correspondence between formal (Lean) statements and the theorems a paper claims, and the provenance of the text. A point is retained only if a reproducible computation, an explicit counterexample or a verifiable source supports it, and only after its mathematical scope has been checked against the paper’s hypotheses. Every point comes with the script, page or source that lets anyone reproduce it. This note describes the principle, the checks and the validation chain, the safeguards for authors, and a pilot on the manuscripts released by OpenAI in October 2026
Authors
- Claire David (ORCID: https://orcid.org/0000-0002-4729-0733)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-19
- DOI
- https://doi.org/10.5281/zenodo.24356845
- Primary Topic
- Mathematics, Computing, and Information Processing
- Type
- article
- Field-Weighted Citation Impact
- 0.00