Nikon Reopens Review of Winning Microscope Video After Scientists Question AI Use

A first-place microscopy video is back under scrutiny, with Nikon re-reviewing its original vetting materials after scientists challenged the footage’s biological accuracy and use of AI. The winning entrant, Ning Xu, says AI did not generate the cilia or their movement. The dispute instead puts the video’s processing—and whether its vivid structures faithfully represent microscope data—at the center of the review.

A prize, then questions about the biology

Nikon Instruments named Xu the winner of its 2026 Small World in Motion competition on September 15. The annual contest celebrates videos made with a light microscope. Xu’s entry shows cilia, tiny hair-like structures lining part of the lungs, moving above a field of red, purple and blue features.

The tissue came from a child with primary ciliary dyskinesia, a rare genetic disorder that causes chronic lung, sinus and ear infections. Nature’s account of the controversy noted that the colored structures beneath the cilia were not described in the competition website’s blog post about the entry.

Days after the award, microscopy experts began questioning the clip on social media. Edward Phelps, a University of Florida bioengineering researcher, focused on the colored features. He said the purple structures resembled mitochondria but appeared outside cells and were as large as cell nuclei—a combination he said does not occur in biology.

Phelps also said the blue features looked like nuclei but did not behave like them, and he could not identify the red structures. In comments quoted by TechRadar, he described elements appearing and disappearing or moving between cells. Those objections concern what the finished video depicts, not simply whether an AI tool was involved.

Xu draws a line around the AI processing

On September 22, Nikon updated its competition blog to disclose that an unsupervised AI model assisted with post-processing. Nikon described the neural-network method as part of a workflow to distinguish and visualize features in grayscale data, producing a more vivid presentation. That disclosure acknowledged AI use without resolving the objections to the resulting biology.

Xu’s explanation separates the recorded movement from the later visualization. Writing on LinkedIn, he said AI was applied afterward to reconstructed grayscale data to distinguish and color structures with similar shapes. He denied that it generated the experimental movie, the cilia or their motion.

He also said his team enhanced the regions below the cilia without making anatomical claims about the rendered features. Xu framed the presentation as an entry in a competition celebrating microscopy’s beauty: the team wanted it to be visually engaging as well as scientifically interesting. His explanation does not identify the colored features as particular biological structures.

Scientific images are not just illustrations

Melanie White, University of Queensland developmental biologist, speaking to Nature

Nikon’s review remains open

Nikon says Xu is cooperating and has supplied detailed documentation covering the microscopy equipment, imaging methods and processing techniques used to create the source video. The company is examining that material alongside the information submitted during initial vetting. As of Friday evening, October 2, a contest spokesperson had no update for Gizmodo.

A separate allegation concerns the video’s digital labeling. Ian Donovan, a PhD student at UT Southwestern Medical Center, claimed to have found an embedded SynthID watermark identifying synthetic or AI-generated content. That remains an attributed claim, not a conclusion announced by Nikon about how the winning entry was made.

Beauty does not settle fidelity

The contest’s boundaries also matter. Fifth-place winner Patrick Hickey told the BBC, in remarks quoted by Gizmodo, that its rules prohibited generative AI from producing imagery and required entries to be captured under a microscope. Nikon has not announced whether Xu’s processing violated those rules.

Markus Sauer, a microscopy researcher at the University of Würzburg, told Nature that AI visualization is not automatically problematic. The problem arises when it misrepresents, exaggerates or alters experimental findings. White said the video’s fidelity to the microscope measurements was unclear, and that the tool might have introduced structures absent from the underlying data.