The paper is real. It was also withdrawn sixteen years ago.

A different kind of failure

Existence checking and semantic support checking both pass here. Correctly titled. Correctly identified. The paper is exactly what it claims to be. What changes the picture is a piece of metadata living alongside the paper rather than inside it: a retraction notice, which most citation tools never think to look for, because they were built to answer a narrower question.

Retraction status, to Merlise, is part of the evidence a source carries, not a footnote to skip past. It gets checked through publisher metadata and the biomedical retraction record, then surfaced in the ledger the moment a citing sentence leans on a source that has one.

Context decides the verdict

Citing a retracted paper is not automatically wrong. A sentence discussing the history of a scientific controversy, or using a retracted study as a documented example of bad evidence, is doing something different from a sentence that leans on the same paper as ordinary support for a current claim. Merlise reads the surrounding sentence before deciding which situation applies, never just the bibliography entry on its own.

A red flag on the citation would tell a writer little. A specific prompt does more: this reference resolves to a retracted paper, and the text should say so if it plans to use the source at all.

The evidence ledger

The same claim by claim view the product shows on a live document, built from this case.

How it resolves

The resolving record