Nikon announced on October 9, 2026 that the original winning entry of its 2026 Small World in Motion microscopy video competition did not comply with the contest rules on generative AI, and revoked the award given to Dr. Ning Xu of Tsinghua University. The original runner-up, Nguyen Nam Nhat of Vietnam, whose video shows a tiny nematode interacting with the single-celled organism Dileptus, moves up to first place, with other rankings shifting accordingly. Nikon stressed that the decision is not a judgement on the entrant's professional standing or scientific contributions, and said it will review its rules and judging procedures.
The first doubts came from an expert with twenty years of watching cilia
The disqualified video claimed to use super-resolution microscopy to show abnormal ciliary motion in the airway of a child with primary ciliary dyskinesia. After the win was announced, several microscopy and cilia researchers objected publicly: some structures appeared or vanished abruptly, some cilia seemed to materialize from nowhere, and sizes did not match real samples. Robert Hirst of the University of Leicester, who has diagnosed this condition by ciliary beat pattern, length and cell shape for about twenty years, said he knew something was wrong on first viewing. The criticism led Nikon to re-examine the footage and supporting material and consult jury members, reversing its initial finding of no violation.
The question is not whether AI was used, but what it touched
Xu had already acknowledged using AI in production, explaining that it distinguished and visualized features in reconstructed greyscale imagery, mainly to improve presentation, and was not used to generate the experimental video itself, the cilia or their motion; he believed this complied with the rules. That is exactly where the dispute sits. In scientific imaging, the line between coloring, enhancement, feature separation and generation is hard to draw from an author's own description, while the contest rules state that entries must be filmed through an optical microscope, that AI-generated videos are not accepted, and that organizers may demand the original footage. This year's competition drew 346 entries from 40 countries, and scientific credibility is the premise of its entire value.
The ranking is fixed; the rules are the harder part
For scientific imaging contests, this case is less a cheating story than an unanswered question. Super-resolution reconstruction, AI denoising and AI segmentation coloring are becoming routine microscopy tools; banning all of them is unrealistic, while allowing all of them leaves judging without a basis. Nikon has said it will review its rules and procedures, and plausible directions include requiring disclosure of AI processing steps, retaining raw data for audit, and defining enhancement versus generation by what was processed. Until such details exist, researchers entering similar contests are safest documenting every AI step in their submission and keeping reproducible original files, because once challenged, the burden of proof effectively falls on the author.
