Nikon has disqualified Dr. Ning Xu’s original first-place video in its Small World in Motion competition, saying the entry did not comply with its rules on generative AI. The decision, reported on October 9, 2026, followed a re-evaluation of the video and supporting materials, including consultation with members of the judging panel. (PetaPixel)
The company explicitly limited its finding to competition eligibility. It said the decision should not be read as a judgment of Xu’s professional reputation, scientific contributions or intent—a distinction that separates the documented rule violation from allegations of scientific misconduct. (PetaPixel)
Xu acknowledged using AI-assisted post-processing but maintained that he believed he had followed the rules, according to the reporting. His explanation places the dispute at the boundary between processing microscope-derived imagery and producing content that a competition considers impermissibly generated. (PetaPixel)
What Nikon reviewed
The entry was described as showing moving cilia in the airway of a child with primary ciliary dyskinesia, or PCD. That is the claimed subject of the footage, not an independently verified account of its biological provenance. Nikon’s own competition post describes the work as a rendition and identifies AI assistance in post-processing. (Nikon Small World via LinkedIn)
Scientists challenged the appearance of the footage in comments on Nikon’s LinkedIn page. Those objections questioned whether the visible structures and their movement accurately represented the biological subject. They are concerns raised by commenters, rather than independent findings established by this article. (Nikon Small World via LinkedIn)
Nikon subsequently reviewed the entry and revised the rankings. Its reported statement identifies noncompliance with generative-AI rules as the reason for disqualification; it does not provide a detailed technical determination of which individual image features were altered or how they were produced. (PetaPixel)
Xu’s explanation and the rule
The Verge reported that Xu described using an unsupervised neural-network method to distinguish and visualize features in reconstructed grayscale images from super-resolution optical imaging. That description is his account of the processing, not an independently reproduced or verified reconstruction of his workflow. (The Verge)
PetaPixel quotes the competition restriction as: “AI-generated videos are not permitted.” The complete rules applicable to this entry, including any definitions or exceptions, could not be directly verified for this draft. The quoted prohibition therefore should not be expanded into a claim that Nikon barred every use of AI in image processing. (PetaPixel)
Taken together, the reported decision and Xu’s explanation establish a disagreement over eligibility, not proof that the entire video was generated from scratch. Calling the entry wholly fabricated would go beyond the scope of Nikon’s stated finding and the evidence available here. (PetaPixel)
A replacement winner
Nguyen Nam Nhat is the replacement first-place winner, according to both Ars Technica and PetaPixel. His video features a roundworm and a single-celled Dileptus. The revised result is corroborated by those reports, although Nikon’s updated results page could not be directly retrieved during preparation of this draft. (Ars Technica)
Other competitions distinguish enhancement from generation
The broader issue is not unique to scientific imagery. World Press Photo’s 2026 entry rules prohibit synthetic images and generative fill, while allowing some AI-powered enhancement when it does not significantly change the image or add or remove information captured by the camera. Those are that organization’s standards, not an interpretation of Nikon’s rules. (World Press Photo)
World Press Photo lists limited denoising, automatic adjustments and selection tools among potentially acceptable uses. Its rules make acceptance depend on what the processing does to the photograph, with the organization and jury determining whether the result remains within the permitted boundaries. An AI label alone does not resolve eligibility under that framework. (World Press Photo)
The National Press Photographers Association’s 2026 Best of Photojournalism guidelines likewise address processing explicitly. They prohibit generative tools that create, add, remove, expand or alter images or videos, while distinguishing ordinary toning and color correction from manipulation. The guidelines also provide for requesting camera-original images when judges have concerns. (National Press Photographers Association)
For scientific communicators and visual-production teams, these comparisons illustrate why the specific processing operation matters alongside the name of the software. A workflow described as enhancement still needs to be assessed against the receiving organization’s rules and the information preserved or changed in the finished image. That is an implication of the policies, not a further finding about Xu’s video. (World Press Photo)
Nikon said the episode demonstrated a need to revisit rules and evaluation procedures for future Small World and Small World in Motion entries. Its stated response therefore extends beyond revising the rankings to reconsidering how future submissions will be assessed. (Ars Technica)
