Transparency notice

AI-generated content disclosure

Content on CopeCheck is generated by artificial intelligence and reviewed for methodology, not fact-checked line-by-line by humans. Model outputs may contain errors.

What is generated by AI

Automated systems collect candidate political and media claims, extract the claim and its attribution, compare it with cited public information, and draft the displayed score, rationale, summary, and classification. Labels on score and article pages identify this material at the point of use.

Model families

CopeCheck currently uses OpenAI Codex with the GPT-5.6 Luna model at maximum reasoning for primary scoring and analysis. MiniMax remains the automatic fallback if the Codex scoring node is unavailable. A separate Codex-assisted publication path can structure items that pass its quality gate. Some historical records may identify earlier model families in their own metadata; the model recorded with an item is authoritative for that item.

Method in brief

1. A source is collected and the relevant claim and speaker attribution are extracted.
2. The claim is compared with public datasets and source evidence relevant to employment, productivity, wages, vacancies, and other structural indicators.
3. An automated rubric produces a score, classification, rationale, confidence, and review flags. Low-confidence or weakly attributed claims are held back or marked for review.
4. The result is published with links to available source material so readers can inspect the evidence themselves.

The score is analytical commentary, not a statement of objective fact and not a substitute for reading the underlying sources.

Corrections

If a claim is misattributed, a source is incomplete, or a score relies on incorrect data, please use the contact and corrections form. Include the page URL and the specific issue so it can be reviewed.