An interactive front-end over the frozen Article 50 public-evidence scorecard.
AI disclosure: parts of this text were artificially generated with AI assistance and reviewed by the author. The model, and the conflict it creates, are named in the Conflict of interest section.
Live: https://nmairesearch.github.io/article50-explorer/ Paper / data: 10.5281/zenodo.21819102 (concept, always latest) · scoreboard repo
Click any cell to read the artefact behind it: the clause, the exact question asked, the verdict, the evidence URL, the access date and the account state it was captured in. Twelve consumer AI products, ten items each, measured on 5 and 6 August 2026, four days after Article 50 of the EU AI Act began to apply.
This is not a compliance assessment. Every question is of one form: is evidence of X publicly available, yes or no. no artefact found is a statement about a bounded search, never about a provider. The 50(7) signatory column is reported beside the table and is never added to the ten-item total.
The frozen rubric does not count a product name containing "AI" as a disclosure ("Ask Meta AI" scores no). The toggle applies the alternative rule and highlights exactly which cells move under it, so a reader who disagrees with the call can see its effect rather than take it on trust.
python3 build_tool.py
Reads article50_scores.csv (the same file deposited on Zenodo) and regenerates index.html. Standard library only, no network, no dependencies. The status encoding is always backed by a glyph shape (✓ / ✗ / ○), so it survives colour-blindness and greyscale print; no is drawn neutral rather than red, because the instrument scores public evidence, not compliance.
The same eleven questions will be asked again shortly after the 50(2) marking transitional deadline on 2 December 2026, with this rubric unchanged. The movement between the two runs is the intended result.
The research question, method, sourcing decisions and analytical judgements are the author's. Parts of this text were artificially generated with the assistance of a model made by Anthropic and were reviewed by the author before publication, and that company's products are in the scored set. The rubric was fixed before any artefact was opened, that product was scored last, every cell relating to it was re-checked, and every quotation from its documentation is verbatim.
This tool renders the frozen file set of version 1, 10.5281/zenodo.21849684. article50_scores.csv here is byte-identical to the deposited copy.
CC BY 4.0.