Every day an independent AI reviewer — a separate model call, never the one that wrote the content — draws random samples from everything published across all language sites in the past 30 days: in-depth Bible studies, daily devotionals and long-form articles. Each sample is scored 1–10 against Faith.network's published editorial standard on six dimensions: language quality, verbatim Scripture quotation (verified against our own Bible database text), biblical grounding (the content is laid side by side with the passage it explains — every claim must flow from what Scripture itself says), theological soundness, tone & vision, and clarity. Every major or critical finding is corrected the same day: the live content is revised against the standard, re-checked by the reviewer, and the finding is published together with its resolution. Any critical finding or a weak daily average also alerts the editorial team immediately by email. Reports are published unedited — including the misses.
This is the permanent record for septembre 2026 (still in progress). Severity spread: 0 clean · 9 minor · 1 major · 0 critical.