The results show that human review remains essential before those transcripts are used to build a criminal case or hold an officer accountable, the researchers said. By pinpointing exactly where AI stumbles, the study helps investigators and researchers know what to check before relying on a transcript \u2014 for example, whether voices are missing or statements are attributed to the wrong speaker.\u00a0<\/p>\n
The software identified fewer participants per transcript in the AI-generated texts? and assigned longer passages to a single speaker, losing some of the back-and-forth. AI identified an average of 3.3 participants compared with 4.8 in the versions edited by people. It maxed out at seven, when its human counterparts identified as many as 16. AI transcripts were also watered down, containing about 25,000 fewer words across the sample. Missing dialogue could mean losing an explanation, warning or response that helps establish why an encounter escalated.<\/p>\n
The researchers compared transcripts from 176 body camera videos<\/a> that generated around 23 hours of footage covering 73 incidents. AI created one set, while human transcribers edited the other.\u00a0<\/p>\n The amount is a mere fraction of the more than 350 hours of footage generated by Kansas City police in a single day, the authors explained.<\/p>\n \u201cRecording an encounter is much easier than finding the time to watch and analyze everything that gets recorded,\u201d Reid said.\u00a0<\/p>\n \u201cWe have turned supervisors into \u2018movie watchers\u2019 when they should be present and focused on developing the skills of their officers,\u201d Shellie Solomon, chief executive officer of Justice & Security Strategies<\/a>, a research and consulting firm, added.<\/p>\n With little time to spare, the prospect of AI turning footage into searchable text is all the more appealing.\u00a0<\/p>\n But getting the words down is only part of the job.<\/p>\n The bots also need to keep track of who said what \u2014 and sometimes, they got that wrong. There were no amphibian cameos among the samples, but one mix-up in particular raised some eyebrows. In a tense exchange between an officer and a civilian, the AI captured nearly all the words. However, it combined statements from different speakers.<\/p>\n \u201cI ain\u2019t selling no more dope,\u201d the person stressed, insisting they weren\u2019t going back to prison.<\/p>\n \u201cYou\u2019re going,\u201d the officer countered.<\/p>\n \u201cI\u2019m done selling PCP,\u201d the person responded.<\/p>\n The AI transcript mistakenly made all three lines come out of the same mouth.<\/p>\n \u201cSomeone \u2026 could come away with the wrong impression,\u201d Reid said.<\/p>\n It\u2019s easy to see where the confusion comes from, she added. Bodycam recordings aren\u2019t made under studio conditions. Amid the chaos, even someone at the scene could struggle to tell speakers apart.<\/p>\n The Kansas Police Department did not immediately respond to a request for comment from Northeastern Global News regarding the findings of the study.\u00a0<\/p>\n In the end, camera footage \u2014 or the AI used to analyze it \u2014 won\u2019t automatically make policing more transparent. However, it provides an independent log of events and draws attention to footage that might otherwise slip through the cracks.<\/p>\n A searchable record means public defenders can \u201cfind the two minutes they\u2019re looking for without having to scrub through a three-hour video file,\u201d Logan Seacrest, criminal justice and civil liberties fellow at the R Street Institute<\/a>, a public policy think tank, told NGN.\u00a0<\/p>\n \u201cThe catch is that AI puts a layer of interpretation between the recording and the record,\u201d he said. And due to a phenomenon called automation bias, people tend to give a machine the benefit of the doubt.<\/p>\n For that reason, it\u2019s important to keep the roles clear.<\/p>\n \u201cThink of (AI) as the investigative assistant who prepares the file,\u201d Solomon explained. \u201cThe humans decide what the details mean.\u201d<\/p>\n Northeastern Global News, Sign up for NGN\u2019s daily newsletter for news, discovery and analysis from around the world.<\/p>\n Reid also suggested checking what it leaves out of the review pile. Otherwise, those encounters may never get a second look.<\/p>\n Missing words aren\u2019t the only potential omission, however.\u00a0<\/p>\n![\"Savannah](\"https:\/\/www.newsbeep.com\/ca\/wp-content\/uploads\/2026\/10\/100226_MM_Katie_Karmelek_009.jpg\")
10\/02\/26 \u2013 BOSTON, MA. \u2013 Savannah Reid, a Ph.D. student in the School of Criminology and Criminal Justice, poses for a portrait on Oct. 02, 2026. Photo by Matthew Modoono\/Northeastern University
Ph.D.<\/p>\n![\"Savannah](\"https:\/\/www.newsbeep.com\/ca\/wp-content\/uploads\/2026\/10\/100226_MM_Katie_Karmelek_029.jpg\")
10\/02\/26 \u2013 BOSTON, MA. \u2013 Savannah Reid, a Ph.D. student in the School of Criminology and Criminal Justice, poses for a portrait on Oct. 02, 2026. Photo by Matthew Modoono\/Northeastern University
Ph.D.<\/p>\n![\"A](\"https:\/\/www.newsbeep.com\/ca\/wp-content\/uploads\/2026\/10\/100226_MM_Katie_Karmelek_005.jpg\")
10\/02\/26 \u2013 BOSTON, MA. \u2013 Savannah Reid, a Ph.D. student in the School of Criminology and Criminal Justice, poses for a portrait on Oct. 02, 2026. Photo by Matthew Modoono\/Northeastern University
Ph.D.
\nCriminal justice Ph.D. student Savannah Reid examines what AI police reports capture \u2014 and lose \u2014 when police encounters become written records. Photos by Matthew Modoono\/Northeastern University<\/p>\n
in your inbox.<\/p>\n