Artificial intelligence may help officers produce clearer reports faster. But prosecutors are beginning to raise a more consequential question:
Who is responsible when AI gets a fact wrong?
For law enforcement, the answer should remain straightforward: the officer.
Michelle Williams, an Assistant District Attorney with the Fulton County District Attorney’s Office in Georgia, offered particularly practical advice when asked what officers should consider when using AI-assisted report writing.
Williams stated:
“PROOFREAD and DOUBLE CHECK. I use an AI assistant for personal life tasks, so I know that AI will make assumptions (or ‘hallucinate’) to fill in gaps in information. If an officer INSISTS on using AI, the only way to catch these assumptions and hallucinations is to proofread and double-check statements about evidence, statutes, and quotes from witnesses.”
That warning deserves attention because police reports occupy an unusually consequential position in the criminal justice system. Reports can influence charging, probable cause determinations, plea negotiations, suppression litigation, and an officer’s later testimony. Legal scholarship has consequently raised concerns about hallucinations, factual gap-filling, bias, and how AI-generated language could affect this foundational record (Ferguson, 2025).
Williams’s recommendation also aligns with emerging law enforcement practice.
The U.S. Department of Justice’s COPS Office reports that agencies implementing AI-assisted reporting have required officers to review generated narratives, complete missing information, and attest to the accuracy of the narratives. Some agencies have added supervisory review and recurring audits comparing reports against body-camera footage (Office of Community Oriented Policing Services [COPS Office], 2025).
But prosecutors are not unanimous about the technology.
Oklahoma City officials consulted district attorneys covering four counties and the U.S. Attorney’s Office before expanding AI-assisted reporting. Some prosecutors were receptive; others wanted to see how defense attorneys would challenge the reports in court. Elsewhere, prosecutors have taken substantially more restrictive positions because of hallucinations, preservation, and evidentiary concerns.
That disagreement suggests an important principle for police chiefs: Don’t wait until an AI-assisted report reaches a prosecutor to ask what the prosecutor thinks about AI-assisted reports.
Bring prosecutors into the conversation before deployment.
Establish what AI may do, what it may never do, what records should be preserved, what disclosures may be necessary, and how officers must verify the final document.
AI can assist with writing, but it cannot assume responsibility for the truth.
That remains human.
Edward Brown of Klyvorek says, “Imagine every officer ending the shift with documentation completed, supervisors correcting fewer reports, and command knowing exactly how AI participated in every document. That’s a good day.”
_______________________________________________________________________
KLYVOREK, a division of the American Academy of Advanced Thinking, LLC, is an AI-assisted law enforcement documentation platform designed to help officers create clear, structured incident reports and draft search and arrest warrant applications while keeping the officer in control of the final document.
_______________________________________________________________________
References
Ferguson, A. G. (2025). Generative suspicion and the risks of AI-assisted police reports. Northwestern University Law Review.
Office of Community Oriented Policing Services. (2025). Using AI to write police reports. U.S. Department of Justice.
