Nikon has disqualified the first-place winner of its annual microscopy video competition after concluding that the submission violated contest guidelines prohibiting the use of generative artificial intelligence, the company confirmed.
The winning entry in the 16th annual Nikon Small World in Motion contest was submitted by Dr. Ning Xu, an optical engineer at the National University of Singapore. Nikon Instruments first unveiled the selection in mid-September, highlighting footage that claimed to capture microscopic, hair-like cilia beating inside airway tissue from a child diagnosed with primary ciliary dyskinesia, a rare respiratory condition, according to the BBC and Nature.
Scientists flag biological errors and watermarks
The decision to disqualify the entry followed swift online scrutiny from scientific researchers after Nikon shared the video on LinkedIn. Observers quickly questioned unusual red, blue and purple forms visible beneath the moving cilia.
Edward Phelps, a bioengineering researcher at the University of Florida in Gainesville, pointed out on LinkedIn that the purple features appeared to mimic mitochondria of an unnatural scale. Phelps wrote that such a formation composed of extracellular mitochondria sharing the size of cell nuclei “does not occur in biology,” while adding that nearby blue shapes behaved abnormally for cell nuclei.
Questions about the video deepened when Ian Donovan, a doctoral student at UT Southwestern Medical Center, discovered that the footage contained a SynthID digital watermark, CNN reported. SynthID, developed by Google’s DeepMind research division, is widely used across the technology sector to identify media generated or altered by AI algorithms.
Entrant acknowledges post-processing tools
In statements posted to LinkedIn, Xu defended the fundamental authenticity of the source material while acknowledging the use of advanced computing methods to enhance the visuals. Xu explained that his team relied on an unsupervised neural-network method during post-processing to distinguish and color features from reconstructed grayscale optical data.
Xu asserted that the baseline recording of the cilia and their physical motion remained authentic, Nature reported. He added that the team rendered features below the cilia “without making anatomical claims about what those rendered features represent.”
Patrick Hickey, who placed fifth in the contest, told the BBC that microscopy rules must remain strict because researchers rarely receive public showcases for their imaging. Hickey noted that competition regulations explicitly prohibited using generative AI to produce contest material and required all submissions to be captured under an actual microscope.
Other scientists expressed concern over how computational manipulation affects research integrity. Melanie White, a developmental biologist at the University of Queensland in Australia, told Nature that scientific imagery serves as direct data, noting researchers “need to be able to trust that what we are seeing is grounded in the underlying measurement.”
Contest rankings updated
Nikon confirmed that Xu’s video failed to meet contest standards governing generative AI and removed the clip from its official website, The Verge reported. Following the disqualification, Nikon updated its competition leaderboard, elevating a submission by Nguyen Nam Nhat to first place.
In a formal statement addressing the decision, Nikon emphasized that stripping the award “should not be interpreted as a judgment of the entrant’s professional reputation, scientific contributions, or intent.”
Looking ahead, Nikon said it will review its eligibility rules and submission evaluation procedures before launching future competitions, though organizers have not yet detailed what specific detection tools or verification steps will be implemented for future entrants.
