Nikon Withdraws Winning Film After SynthID Finding
Nikon is withdrawing the first-place award in Small World in Motion after weeks of controversy: researchers spotted biologically impossible structures, and a medical student is reported to have flagged a SynthID watermark in the film via Google's Gemini, according to PetaPixel and CNN as reported by Gizmodo. Nguyen Nam Nhat of Vietnam moves up to first place.
Nikon announced on Friday, October 9, 2026, that it is withdrawing this year's first prize in Small World in Motion, the video category of the microscopy competition, after concluding that the winning film violated the competition's rules on generative AI. The $3,000 prize instead goes to previous runner-up Nguyen Nam Nhat of Vietnam, according to DPReview (97ba6103).
The decision came after weeks of mounting pressure. The results were announced on September 16 (e3fad9ad), and the winning film, made by Ning Xu, was covered by a range of media outlets before subject-matter experts began to react.
A Film That Did Not Look Like Biology
The winning film showed cilia in airway samples from a child with primary ciliary dyskinesia (PCD), a rare hereditary disease that affects ciliary function. According to Gizmodo, experts noticed that the film was full of structures that do not exist in biology, and that it contained a SynthID watermark (a199f15a).
Robert Hirst, a researcher at the University of Leicester with 20 years of experience in PCD diagnostics, told the BBC that he diagnoses based on wave form, length, cell size and cell shape. What the film showed, he said, "[resembles] nothing from PCD patients," adding that the case has had a negative impact on PCD patients, physicians and support networks worldwide (as reported via Gizmodo).
Exposed Outside the Judging Panel
Neither Nikon's judging panel nor the company's own checks caught the problem. It was Ian Donovan, an MD/PhD student at UT Southwestern Medical Center, who, according to PetaPixel and CNN as reported by Gizmodo, ran the film through Google's Gemini. The tool flagged a SynthID watermark — the invisible mark Google DeepMind develops to label AI-generated content (c013c798).
Nikon's own response shifted along the way. Early in the case, the company told the BBC: "At this time, we don't see that any rules were violated" (as reported via Martin Cid Magazine). Only on October 9 did Nikon state on its website: "After thoroughly re-evaluating the video and supporting materials and consulting with members of our judging panel, Nikon has determined that the winning video did not comply with the competition rules regarding generative AI" (e3fad9ad, Ars Technica).
Xu: AI for Post-Processing, Not Generation
Xu has consistently denied that the film was AI-generated. According to Ars Technica, he said AI was used "to distinguish and visualize features in the reconstructed grayscale images," but rejected that AI was used "to generate the experimental movie, the cilia, or their motion" (153d1633). In a blog post on Nikon's website on September 22, he wrote that AI was used for video post-processing in accordance with the competition rules, and that he was cooperating with Nikon's investigation (as reported via CHOSUNBIZ).
The rules nevertheless stated explicitly that "A.I. generated videos are not permitted and Sponsor holds the right to request the original video for verification," according to The New York Times. The line between AI-based post-processing and "AI-generated" content, however, is not clarified in the sources — and Nikon has not taken a position on that question. Spokesperson Remi Yuter wrote that the decision rests solely on the eligibility rules and "should not be interpreted as a judgment of the entrant's professional reputation, scientific contributions, or intent" (as reported via Gizmodo).
A New Prize List
After the disqualification, Nikon has moved up the winner list (153d1633):
- Nguyen Nam Nhat (Vietnam) — film of a roundworm and the single-celled Dileptus
- Benedikt Pleyer (Bavaria, Germany) — jellyfish larvae suspended in water droplets
- Andrew Moore (Howard Hughes Medical Institute, USA) — synchronized cell division in neighboring cells
This year's video competition received 347 entries from 40 countries, and the contest named winners from first through fifth place as well as 26 honorable mentions (as reported via CHOSUNBIZ).
The Rules Must Be Reassessed
Nikon acknowledges that the case reveals a structural problem. Its statement reads: "Advances in imaging and artificial intelligence continue to evolve rapidly, creating new challenges for organizations across many fields. This event has shown the need for Nikon Small World and Nikon Small World in Motion to revisit rules and evaluation procedures for future competition entries" (as reported via Ars Technica).
What that means in practice has not been announced — the sources confirm only that the rules and evaluation procedures will be revisited, not how.
The scientific community points out that the problem may not be the AI use itself. Aydogan Ozcan, professor of electrical and computer engineering at UCLA, told The New York Times that "AI use in scientific video is not new," and that AI helps researchers improve resolution and contrast in microscope images. The problem, he said, is the lack of transparency about the AI use (as reported via CHOSUNBIZ).
Unresolved Questions
Several questions remain open based on the available sources. It is not documented how the film passed initial review, or whether the judges ever saw the raw footage. Nor has it been concluded whether Xu's characterization of AI as "post-processing" was actually a rule violation — Nikon judged the case on eligibility rules without addressing intent, and Xu claims he followed the rules.
The external facts are also disputed in the source picture: The New York Times gives September 16 as the announcement date, while CHOSUNBIZ writes September 15 — and the sources disagree about Xu's affiliated institution, with some media outlets naming the National University of Singapore and others Tsinghua University. What is correct cannot be determined from the available material.
What the case demonstrates is that competitions and research communication now lack a shared definition of where AI enhancement ends and AI fabrication begins — and that this time the decision was made on the basis of a watermark almost no one saw, detected by a single person with the right tool.