A check I did: the 18,000 antibody images

This is the shape of what you get, on a claim that was already public.

The claim as it gets repeated

“More than 18,000 questionable images found in antibody catalogues of 15 companies.”

That is the headline of a Nature news piece, 25 August 2026, by Dan Garisto: https://www.nature.com/articles/d41586-026-02635-w

What I could reach

• The headline and the opening lines are open to anyone. The piece reports that thousands of antibodies sold by multiple research vendors have catalogue images that “seem to have been digitally altered,” according to a scientific-integrity specialist. Both of those phrases are in the lede at the link above.
• Nature also ran a companion piece naming one supplier, reporting more than 100 suspicious images in that company's antibody catalogue.

What I could not reach

• The body of the article is subscription-only. The per-company breakdown, the identity of the specialist's dataset, and the suppliers' full responses are not in front of me.
• Which means: anyone quoting the 18,000 number is quoting a count they most likely cannot open.

What holds, what does not, what stays open

• Holds: the number, the 15-company scope, and the fact that the finding is attributed to a scientific-integrity specialist rather than to a journal or a regulator.
• Does not hold, if you push it: “18,000 fake images,” or “15 companies faked their data.” The word in the record is questionable, and the allegation is about catalogue images, not about the underlying experiments.
• Stays open: how many of the flagged images were corrected, and what the suppliers say. The public record here is a news report. It is not a retraction.

Why this is the job

The headline is true. The conclusion most people would draw from it is not, and the source is paywalled, so almost nobody checks. That gap is what I get paid to stand in.

Want one claim checked? [email protected]

James. Research with receipts. [email protected]