Nikon Disqualifies Small World in Motion Winner Over Generative AI Use

Nikon has disqualified the video that originally won first place in its Small World in Motion contest because it "did not comply with the competition rules regarding generative AI," the company said in a statement.

The winning entry came from Dr. Ning Xu. According to the BBC, Xu's video claimed to show "tiny, hair-like structures called cilia moving in the airway of a child with the respiratory condition PCD." PCD, or primary ciliary dyskinesia, is a rare inherited condition in which the cilia that clear mucus from the airways do not move properly — a subject where authentic microscopic footage carries real diagnostic weight.

The disqualification caps a controversy that played out in public over roughly two weeks. Nikon confirmed last week, on October 1, that it was reviewing the winning video after skepticism about its authenticity spread online. Researchers and microscopy enthusiasts had raised doubts about whether the footage showed what it claimed. Nikon has now concluded that it did not meet the competition's standards.

What did Dr. Xu admit?

Dr. Xu responded to the accusations directly. In a comment on LinkedIn, he admitted to using AI on the video: "An unsupervised neural-network method was subsequently used for AI-assisted post-processing to distinguish and visualize f…" The comment was truncated in the available source material, but the admission is explicit: a neural network was applied to the footage after capture.

That admission sits at the center of the dispute. The question for the contest, and increasingly for scientific imaging broadly, is where the line falls between traditional post-processing — contrast adjustment, deconvolution, colorization — and generative AI that can synthesize or alter structures that the microscope never actually recorded.

Nikon's ruling gives one answer: an "unsupervised neural-network method" applied in post-processing falls on the wrong side of the line for this competition.

Why does one contest disqualification matter?

The stakes extend beyond a single prize. Nikon's Small World competitions, running for decades, are among the most visible showcases of scientific microscopy in the world. Their credibility rests on the assumption that the images and videos depict real specimens. A first-place winner using generative AI undermines that trust — and Nikon's decision to revoke the award rather than quietly move on signals how contest organizers now have to police AI use explicitly.

The case also lands amid a broader reckoning over AI in scientific imagery and research output. As neural-network tools for denoising, super-resolution, and generation become standard parts of the imaging workflow, journals, competitions, and institutions are being forced to draw enforceable boundaries between enhancement and fabrication. Nikon's disqualification is one of the clearest examples yet of a major scientific imaging body enforcing such a line, with a named winner and a public retraction of the result.

The episode follows a familiar pattern now seen across photography, illustration, and academic publishing: online communities spot anomalies, pressure mounts, and the awarding body re-reviews. Nikon acted after the initial wave of skepticism, reviewing the video and ultimately stripping it of first place.

What happens to the contest results?

The video will not stand as the winner of the 2026 Small World in Motion contest. Nikon's statement confirms the disqualification, though the source material does not specify which entry, if any, takes over the first-place position.

For Dr. Xu, the LinkedIn comment stands as his own account of what happened: AI-assisted post-processing was used, and Nikon judged that this violated the rules. For the microscopy community, the precedent is set — at least for this competition, generative AI in post-processing is disqualifying, even when the underlying footage may be real.

Expect other scientific imaging competitions and journals to watch how Nikon's rule is worded and enforced, as they face the same question of what counts as a real image in an era when neural networks can finish the picture.