Evaluation type Beta
Content Credibility
Evaluates whether a page is readable, attributable, sourced, specific, original, and honestly maintained. It does not claim to prove whether a human or an AI wrote the page.
Powered by AIOTruth. Beta.
Is this page credible enough for an AI system to quote it?
Purpose
Evaluates whether a page is readable, attributable, sourced, specific, original, and honestly maintained. It does not claim to prove whether a human or an AI wrote the page.
How it runs
| Inputs | A single public page URL. |
| Retrieval method | AIOInsights fetches the page as served and reads its text, structure, links, and attribution. Only the public page is read. |
| Observable signals | Readable, Attributable to a named author or organization, Sourced, Specific rather than generic, Original rather than templated, Honestly maintained, with visible dates and corrections. |
| Weighting | Content Credibility scores across a set of pillars, published in the live methodology on AIOInsights. It is a beta type, and the rubric is still being refined. See the live methodology on AIOInsights. |
| Not applicable logic | A pillar that cannot apply to a page is marked not applicable rather than scored against the page. |
| Failure states | A page that cannot be fetched or read produces a stated failure, not a guessed result. |
| Evidence shown | Each pillar result points to what on the page produced it. |
| Result | A Content Credibility result, in beta. The score name and rubric may change as the type matures. |
| Version | Engine v2.4.0. Change history. |
| Data retention | Handled under the AIOInsights privacy policy. AIOInsights privacy. |
| Security and abuse controls | The same request guard and no-execution rules as every AIOTruth type. Rate limited. |
Known limitations
- This is a beta evaluation type and its result should be read as directional.
- Credibility signals are correlated with, not proof of, trustworthiness.
What this does not measure
It does not claim to prove whether a human or an AI wrote the page. That is not something this evaluation, or any honest evaluation, can determine from the text alone.
Run it
The Content Credibility evaluation runs on AIOInsights as the Content Credibility Check. When it surfaces work that needs doing, Digilu implements it.