This section contains the nature of the case and procedural background.

Puloka (D) was charged with murder. D sought to introduce a cell phone video of the fight that had occurred. D claimed the video would show him trying to de-escalate and had to shoot in self-defense. The video shows that D accidently killed the assailant and two bystanders. The video offered into evidence was AI-enhanced. P moved to exclude the video under Frye. The use of artificial intelligence (AI) tools to enhance video introduced in a criminal trial is a novel technique. D intends to admit an AI-enhanced version of a video recorded by a civilian witness on an iPhone. D's expert witness Brian Racherbaeumer was a self-identified videographer and filmmaker who started working with video in 1993. He was very candid and open about the fact that he is not - and has not claimed to be - a forensic video technician and has not been forensically trained. D's witness presented Exhibit 2 as an enhanced version of Exhibit 1 which he said added clarity through the use of an AI video-editing tool in the program 'Topaz Labs AI', and was further processed on Adobe Premier Pro. Topaz adds sharpness, definition, and smoother edges to objects in the video, whereas the source video contained fuzzier images with 'blocky' edge patterns. Racherbaeumer was unable to say whether the Topaz is currently utilized by the forensic video analysis community. Racherbaeumer did not know what videos the AI-enhancement models are 'trained' on, did not know whether such models employ 'generative AI' in their algorithms, and agreed that such algorithms are opaque and proprietary. D's other retained expert, Matt Nodel, could utilize the source video (Exhibit 1) as the basis for his expert testimony. P’s expert witness, Grant Fredericks, is a Certified Forensic Video Analyst, with national and international forensic video analysis credentials. Fredericks testified that about half of his 300 appearances testifying in Court over the last ten years have been for the State or Plaintiff, and about half for the Defense. Fredericks' focused on 'image integrity', and not on creating a smoother, more attractive product for a user. Fredericks stated that Topaz added approximately sixteen times the number of pixels, compared to the number of pixels in the original images to enhance each video frame, utilizing an algorithm and enhancement method unknown to and unreviewed by any forensic video expert. Fredericks explained, that after modification objects in the enhanced video on Exhibit 2 did not maintain their original shape and color from the source video on Exhibit 1. He stated that the proffered AI-enhanced video removed artifacts on individual images, altered shapes, and removed the opportunity to forensically analyze which frames in the video utilized reference, predictive, and bi-directional images. Topaz made proper, accepted forensic analysis of the video impossible. Topaz was in contrast to approved image enlargement techniques like 'nearest neighbor', 'bi-cubic', and 'bi-linear' - which have been utilized by the forensic video analysis community for decades, and which create video products that are reproducible across approved video processing programs such as Adobe Premier Pro, Adobe Photoshop, Amped5, and Axon Investigate. Fredericks referenced the Scientific Working Group on Digital Evidence (SWGDE), and consistent with the SWGDE up-scaling an image (resizing) is referred to as 'interpolation' and that the most accurate interpolation method that preserves image detail when observing small objects is called Nearest Neighbor. Nearest Neighbor helps mitigate the distortion of the object's shape, length, or size represented by a few pixels. The Nearest Neighbor algorithm can reduce the potential of providing a misleading representation of the level of pixel detail in the original imagery. Topaz's findings cannot be evaluated by the defense expert, the forensic video analysis community, or by this court. SWGDE has found that, for 'machine-learning' interpolation algorithms, 'it can be challenging to identify what processes were applied to the imagery and replicate those steps with accuracy.'

The legal issues presented in this case will be displayed here.

The court's holding and decision will be displayed here.

Legal analysis from Dean's Law Dictionary will be displayed here.