What is circular reporting?
Circular reporting occurs when publications repeat or cite the same underlying claim and the repetition is mistaken for independent confirmation.
Several visible links may therefore lead back to one unverified origin rather than several separate observations.
Three websites can repeat one unsupported number
Imagine site A publishes an unchecked number. Sites B and C repeat it. A then cites B and C as supporting evidence. Readers can now see three websites agreeing, although no new independent evidence has been added.
Noah Tavlin's TED-Ed explanation uses circular reporting to describe this kind of propagation. Agreement between pages is useful only after their evidence chains have been examined.
A real case: Guttenberg's invented Wilhelm
On 8 February 2009, an anonymous editor added the invented name Wilhelm to Karl-Theodor zu Guttenberg's German Wikipedia entry. It appears in the archived revision from that date. The editor's first-person account on BILDblog describes the insertion and identifies SPIEGEL among the outlets that adopted the name.
A 9 February revision diff then shows a SPIEGEL report being added as a footnote after the name containing Wilhelm. The claim had returned to the Wikipedia entry with a media citation, without acquiring an independent check of the name.
On 11 February, SPIEGEL published a correction, acknowledging that its name check had relied solely on Wikipedia.
This documented loop illustrates the problem with counting citations as separate evidence. It is not a finding that all Wikipedia entries or news reports are unreliable.
Trace the data, not just the number of links
Ask whether each writer obtained original records or merely quoted another article. Follow the chain to the earliest accessible material, then check whether it supports the current claim.
Matching conclusions can come from genuinely independent observations. They can also result from copying. The links alone cannot distinguish the two.
The following checklist draws on Mike Caulfield's SIFT approach to investigating sources and tracing claims to their original context.
- Record the exact claim, the page URL and its publication or update date.
- Follow each article to the material it cites. Look for original records, data or firsthand reporting.
- Draw the links between sources. Distinguish separate observations from retellings or analyses that reuse the same data.
- Check the original context and later corrections. If a link cannot be traced, mark the claim unverified; a broken chain alone does not prove it false.
Circular citation or independent corroboration?
| Check | Circular citation | Independent corroboration |
|---|---|---|
| Starting material | Several articles lead back to the same account or dataset. | Sources draw on separately obtained records or observations. |
| New evidence | More retellings add no independent check. | A separate observation or source provides another check. |
| Limits | Repetition does not mean the claim is false. | Independence does not guarantee accuracy; methods and context still need checking. |
Why it matters in AI search
AI search can retrieve several retellings of one claim. An answer with more citations has not necessarily acquired more independent evidence.
The AI-source review uses this distinction to discuss repeated retrieval and citation of generated material. It does not classify all AI-written content as false. The task remains to inspect the claim and its original support.
Sources and scope
- TED-Ed, Noah Tavlin: How false news can spread
- Wikipedia revision, 8 February 2009
- BILDblog: the editor's account, 10 February 2009
- Wikipedia citation diff, 9 February 2009
- SPIEGEL: correction, 11 February 2009
- Mike Caulfield: SIFT (The Four Moves), 19 June 2019
The definition source was checked on 2026-10-24. The historical revisions, case sources and SIFT were checked on 2026-11-05. This is a source-based historical example, not an LS-Labs experiment.