Transparency notice

How CopeCheck Codex uses AI

Oracle scores, verdicts and analytical summaries are generated by artificial intelligence. They are automated analytical outputs, not independent human fact-checks.

Models

The main scoring path uses OpenAI Codex with gpt-5.6-luna at maximum reasoning. MiniMax models, including MiniMax-M3 and M2.7, remain in specialist and fallback pipelines.

Method

The system extracts an attributable claim, compares it with linked public evidence and applies the CopeCheck structural tests: productive participation, unit-cost collision, interface collapse, propagation blindness and coordination feasibility. Attribution, confidence and schema checks gate publication.

Limits and corrections

Models may omit context, make unsupported inferences or assign an incorrect score. Verify every result against its linked source. Send corrections with the page URL and supporting evidence to [email protected].