Editorial Policy

Who our content is for

We write for people making real decisions about AI humanizers and AI detectors: students dealing with detector-flagged work, professionals whose writing gets misread as AI, and developers evaluating humanization APIs. The goal of every article is that a reader can act on it — not that it ranks.

How articles are produced

All articles are written and reviewed in-house by the Wibble team, which publishes anonymously. We do not accept guest posts, sponsored placements, or affiliate arrangements, and no third party pays for coverage or rankings on this site.

Disclosure: we review products we compete with

Wibble makes an AI humanizer, and our articles review and compare other AI humanizers. That is a conflict of interest, and we handle it by disclosing it — in this policy and in the articles themselves — rather than pretending to be a neutral review site. Where comparisons rest on testable claims, they are governed by our testing methodology, which requires publishing raw data readers can re-check, including results where Wibble loses.

How competitor facts are sourced

Factual statements about other products — features, limits, pricing — are sourced from the vendor's own published pages (pricing pages, documentation, changelogs, press releases), and we record the date we accessed them. Pricing is verified at the time of writing and may change afterward; if a price matters to a decision, check the vendor's page before relying on it. We do not attribute fabricated output samples or invented behavior to named competitors.

No test, no claim

This is the rule the rest of the policy hangs on. We do not invent detector scores, pass rates, or "we tested N tools" figures. A tested claim — any specific number about how a tool performs against a detector — may only appear on this site when it is backed by published, reproducible data: dated runs, named detector versions, and raw inputs and outputs. Until such data is published, articles say so plainly rather than filling the gap with plausible-sounding numbers. Claims about Wibble itself are held to the same standard.

Reporting an error

If anything we publish is wrong — a stale price, a mislabeled detector capability, a broken claim — email support@wibbleai.com with a link to the page. Verified errors are fixed and noted under our corrections policy.

Related policies

• About Wibble — who we are and why we publish anonymously.
• Testing methodology — how tested claims earn their way onto the site.
• Corrections & updates — how errors are handled once published.