aceModerate evidence
Published recordCrimeProblem Space
More
Public perceptions of police use of artificial intelligence to assist in generating police reports: findings from an experimental study
- Truvace record
- TRV-2026-2663raw data (JSON)
- Version
- v1Published
- First published
- Evidence type
- Survey
- Sources
- 1 sourcePeer-reviewed
- sha256
- d1d8a561fe5d…check it
In the context of policing, artificial intelligence (AI) and large language models (LLM) tools are being touted for their ability to reduce the amount of time officers spend writing reports while simultaneously improving their accuracy and completeness. Concerns have been raised, however, about bias, privacy, legality, inaccuracies, transparency, and the overall impact of the use of AI when generating police reports on legal proceedings and police legitimacy. Despite continued advancements in the technology and…
The problem
A survey experiment of US residents found only low to moderate public support for police using AI to write reports, with support rising only when people were told the tools were useful and how agencies check their accuracy.
What this doesn’t fix
The study used a non-probability sample of U.S. residents, so its findings may not represent the wider US public.
Sources
1 on file- Policing and Societydoi:10.1080/10439463.2026.2675429
Cite this record
v1Truvace Impact Record TRV-2026-2663, v1: “Public perceptions of police use of artificial intelligence to assist in generating police reports: findings from an experimental study.” Truvace, 2026-11-12. /record/TRV-2026-2663 (accessed on the day you copy it). sha256 d1d8a561fe5d7fb9…
Change history
1 versionEvery change made to this record since it was first published, and why. Nothing has changed since it went up.
Certified into the record
Check this record hasn’t been altered
How to check it yourself, without trusting this page
Download any version from /api/record/TRV-2026-2663 and hash its canonical field yourself, for example with shasum -a 256. You should get its content_hash. Each version’s text also ends with prev:and the hash of the version before it (version 1 points to 64 zeros). If anyone had changed a single character of any version after it was saved, those hashes wouldn’t line up.