First-success case study

From one suspicious citation claim to a successful FactSentinel source check.

This public-safe walkthrough uses the South Africa AI policy fake-sources story to show the path a new visitor should complete: pick one concrete claim, inspect the source trail, then install from `/download` for the next in-browser check.

Published May 4, 2026 - Sources checked against public reporting and the official Government Gazette on May 4, 2026

The concrete claim

A reader sees public reporting that South Africa withdrew its draft national AI policy after fictitious references appeared in the policy's source list. Reuters, via Polity, reported that the draft was withdrawn after revelations about fictitious sources, while MyBroadband reported several examples of references that were alleged not to exist.

The first-success check should not try to review the whole policy at once. It should start with one concrete, source-checkable claim: a named citation in the policy was alleged to be fake or unsupported.

Public-safe example: this page is a walkthrough using public sources, not a raw customer activation log.

The first-success path

1. Preserve the claim

Select the exact sentence or citation claim from a news article or draft source list. Keep names, title, year, publication, and page context intact.

2. Run FactSentinel

Use the web checker or extension workflow to review the selected claim, source links, reasoning, confidence, caveats, and model agreement.

3. Inspect the evidence

Open the underlying source trail. A successful first check should make it obvious what still needs manual verification.

What success looks like

The goal is not to declare a whole policy true or false. The successful activation moment is narrower: the visitor learns how to move from a single suspicious citation claim to an evidence trail they can inspect.

Useful output

  • The exact claim stays visible.
  • Sources and caveats are attached to the answer.
  • Model agreement or disagreement is visible.

Manual follow-up

  • Open the official draft policy PDF.
  • Search for the named citation.
  • Compare public reporting against the policy reference list.

Activation signal

  • The user completed a meaningful check.
  • The result created a next action.
  • The workflow is worth installing for future pages.

Why this is a good first check

Fake-source claims are concrete enough for a first success because they have named evidence objects: article headlines, source lists, publication names, authors, dates, and official documents. A reader can check whether the source trail points to real material instead of relying on a broad opinion about AI policy.

The official Government Gazette shows the draft policy existed and was published for public comment. Public reporting then described the withdrawal and the alleged fake-reference problem. That gives the visitor enough public material to practice source inspection without exposing private user logs.

Sources checked

Try the next check in the browser.

Open the download page, install the Chrome extension, and run your first-success check on a live claim from the page you are already reading.