AI content policy

AI tools assist the Certuvo Editorial Team in three ways:

• Drafting. Producing first drafts and revisions of editorial prose, working from source material and briefs supplied by the team.
• Research organisation. Structuring and summarising official publications — blueprints, syllabi, handbooks — so that human editors can work through them systematically. The official documents themselves, not AI summaries of them, remain the cited sources.
• Consistency checks. Flagging terminology drift, broken cross-references and contradictions across the library for human editors to resolve.

No AI output is published as-is. Every page, whatever mix of human and AI work produced its draft, passes the same three human-governed review stages — technical, subject-matter and claims review — described in the content review process, and is approved by a human through the CMS workflow. Responsibility for published content rests with the Certuvo Editorial Team and Editorial Board, not with a tool. There is no separate, lighter review track for AI-assisted material.

The known failure mode of drafting with AI is confident fabrication: plausible statistics, invented details, misattributed facts. Our defence is structural rather than hopeful. Under the editorial standards, every factual claim must carry a stored official source, and the claims review stage checks each one line by line. A claim that entered a draft without a source — whether typed by a person or generated by a model — does not survive claims review. We treat AI-assisted drafts as unverified until that check has passed.

The following uses are prohibited outright:

• Inventing sources, citations or quotations, or attributing statements to official bodies that they did not publish.
• Creating fake people: authors, reviewers, testimonial-givers, or their credentials, biographies or photographs.
• Fabricating statistics, survey results, pass rates or research data of any kind. This prohibition is absolute and is restated in our research methodology.

These are not judgement calls; a page found to contain any of them is unpublished and corrected under the corrections policy.

We disclose at the level we believe is honest and useful: this policy states openly that AI assistance is used in drafting across our editorial content, under the controls described here, rather than implying that some pages are untouched by it. Attribution on each page reflects accountability — the Certuvo Editorial Team as author and the Certuvo Editorial Board as reviewer — because accountability, not keystroke provenance, is what readers need to evaluate. As named individual attribution arrives through our contributor verification programme, this policy will be updated, and the change will be recorded in the page's change history.

If you believe any page contains an error — AI-related or otherwise — the corrections policy is the route: reports enter review within five business days, and accepted corrections are disclosed on the page and in the public changelog. An AI origin is never an excuse; the published page is ours, and so is the error.