Increasing Officer Safety
Stuart Cameron, chief of police in Old Westbury, N.Y., tells A&E Crime + Investigation that AI integration with his department’s body cameras not only cuts down on time-intensive paperwork, but it also keeps officers safe: Instead of writing reports on the side of the road in their cruisers, officers can be better aware of their surroundings and review reports away from the scene.
He emphasizes that officers are required to review and add to these AI-generated drafts, which are just that— drafts. Officers are still required to input certain information about an interaction and ensure the report is accurate and comprehensive.
Cameron says the live translation service is especially useful, because even a couple minutes can make a difference in an active situation. Before the body camera translation was available in his department, calling an interpreter on a live line took a few minutes and was often hindered if an officer couldn’t figure out what language was being spoken. If a child can’t breathe or a person is holding a gun in a back room when officers arrive, a brief delay in interpretation can have major consequences. The AI translation can automatically detect the language being spoken and translate. “It’s almost flawless, and [it can translate] over 50 languages,” Cameron says.
Policing handbooks can also be hundreds of pages long, which can prove difficult for any officer to digest. The body cameras can now respond to verbal queries in real time, which has proved handy in the field as well.
Building Public Trust
In addition to helping generate reports, AI integration in body cameras could help increase public trust in policing by banking examples of police deescalation tactics on calls, says Jillian Snider, a former police officer who is now a resident senior fellow at think tank R Street Institute. She co-authored a July 2025 report on AI and body cameras.
Some of the AI tools on the market can recognize de-escalation tactics in real time and reward officers, which Snider tells A&E Crime + Investigation “incentivizes officers to use de-escalation attempts more.”
“When you have an AI tool that is looking to find positivity in interactions, it's actually something that more officers would be willing to wear without hesitancy or reluctance,” she says.
If there is a complaint against an officer, a deep repository of body camera footage analyzed by AI can also aid police departments in determining whether there is a pattern in an officer’s behavior, rather than relying on reports with limited evidence or perspectives.
Snider says that these AI tools aren’t flawless when it comes to concerns about bias and profiling. For example, the software could tag an interaction with cursing as hostile or violent when the officer or civilian may have been speaking in familiarity or jest. That’s why it’s important that a human reviews all reports the AI generates, she says.
AI can also fail to detect tonal shifts or significant body language leading to an unrepresentative report, says Logan Seacrest, a resident fellow in criminal justice and civil liberties at the R Street Institute and another author of the report.
“AI can also struggle to detect sarcasm, which relies on subtle tonal shifts that are not obvious when reviewing a transcript in isolation," he tells A&E Crime + Investigation. "An officer making a sarcastic comment to a partner may be flagged as a policy violation, or a suspect's sarcastic compliance might be misread as full cooperation.”
Perhaps more problematic than misinterpretations are AI hallucinations in police reports. In tests of generative AI to draft police reports, systems have been caught inventing legal justifications to fill gaps.
“If an AI drafts a report that includes a probable cause, threat or weapon that was not actually there, and an officer signs off on it without catching the error, that false information becomes part of the official legal record,” Seacrest says.
Regulation Lags
Advancements in AI occur at rapid speed, and according to police and policy experts interviewed for this story, no formal governing or regulatory body has yet been established. A few states have passed laws regulating the use of AI in policing, like California’s newly passed Senate Bill 524 that requires a disclosure at the end of a report if it was generated using AI as well as an affidavit by an officer guaranteeing they reviewed the report for accuracy and completion.
“We actually welcome it. I think that for a lot of police chiefs, the more conservative ones, they kind of sit back and they say, ‘OK, well, we'll wait on this. We'll see how this pans out.’ So having the governor and the legislature speak up about this and make their views known to make a law about it, that helps a lot,” Francis says.
Seacrest says that law enforcement leaders and policymakers need to maintain a "human in the loop," emphasizing the AI as a tool rather than a replacement. As it stands, we are behind in developing regulations around AI in the criminal justice system, which follows suit historically as new technology is deployed in the field before we collectively think about its implications on society. While Seacrest believes this technology has enormous potential to improve public safety, safety cannot come at the cost of our civil liberties.
“We cannot outsource final judgments on arrests or use-of-force to a computer," he says. "In short: AI must assist officers, not replace them."
