At the 2026 International Association of Chiefs of Police conference, police executives heard a warning about AI-assisted report writing: The software may generate the report, but the officer will be the one defending it in court.
As Police1 reported in its coverage of the conference, Eric Atstupenas, general counsel to the Massachusetts Chiefs of Police Association, urged agency leaders to look beyond the promise of reduced paperwork. His presentation addressed accuracy, officer credibility, discovery obligations, records retention, and the role of prosecutors in decisions about adopting AI report-writing tools. He also encouraged executives to challenge vendors about data ownership, audit capabilities, system settings, and what happens when an agency ends its relationship with a technology provider.
For companies developing and marketing AI report-writing products, the presentation offers insight into the questions prospective customers are being encouraged to ask. Some vendors may already address these concerns extensively in product demonstrations, technical documentation, and implementation support. Others may have capabilities that receive relatively little attention in their broader marketing.
The presentation offers an opportunity to consider how well vendor messaging reflects the decisions agencies must make. Speed may be a compelling selling point, but it tells only part of the story when agencies are responsible for the report long after it is written.
Efficiency matters, but how is it measured?
Reducing administrative work is a legitimate goal for law enforcement agencies. Officers spend considerable time documenting incidents, and AI-assisted report writing offers the possibility of reducing some of that burden. A product that turns recorded audio, officer notes, or other source material into a structured narrative could make the initial drafting process considerably faster.
But generating a draft is only one part of producing a report. An officer still needs to review the narrative, compare it against the source material, identify inaccuracies, make corrections, and verify that the final report accurately reflects the incident. The officer remains responsible for the report's contents, regardless of how the initial narrative was produced.
Research on AI-assisted police report writing illustrates why efficiency claims require careful measurement.
A 2024 randomized controlled trial published in the Journal of Experimental Criminology examined AI-assisted report writing among 85 officers and 755 reports. Researchers found no statistically significant reduction in report-writing time.
A 2025 follow-up study examined officer and supervisor perceptions. Although 48% of officers assigned to use the AI tool reported perceived time savings, researchers found no statistically significant differences between treatment and control groups in perceived utility, speed improvement, or report quality. Supervisors reported improvements in report quality, completeness, and writing efficiency.
Neither study establishes that AI-assisted report writing cannot improve efficiency. Both involved a particular product in a single agency. Together, however, they illustrate the importance of distinguishing between draft generation speed, perceived time savings, and measured improvements across the complete reporting process.
For vendors, that distinction matters when developing and supporting efficiency claims. Prospective customers need to understand whether reported savings reflect initial drafting or the entire process, including review and correction. They also need to know how those savings were measured, whether through controlled testing, agency deployments, officer estimates, or another method.
Clear explanations of what the technology demonstrably improves can make efficiency claims more credible and useful to buyers.
The legal environment varies across jurisdictions
One of the most consequential points in the Police1 article concerns the role of prosecutors. Atstupenas warned police executives that prosecutorial expectations may develop faster than appellate case law governing AI-assisted reports, and the article describes several examples of how those expectations are affecting agency decisions.
In September 2024, the King County Prosecuting Attorney's Office in Washington informed law enforcement agencies that it would not accept police report narratives produced with AI assistance. Connecticut prosecutors have also imposed restrictions on AI programs used to draft criminal reports.
States are also beginning to establish statutory requirements. California Penal Code Section 13663, effective January 1, 2026, addresses agency policies, disclosure of AI involvement, officer verification, and preservation of the initial AI-generated draft and an audit trail. Utah has established requirements concerning generative AI use in law enforcement records, including disclosure and author certification.
These examples illustrate the different legal and prosecutorial environments in which AI report-writing vendors operate. An agency may be interested in a product's capabilities but face restrictions or requirements that affect whether and how the technology can be deployed.
That doesn't make vendors responsible for determining whether a report is admissible in court. Admissibility remains a legal question, and prosecutorial practices and disclosure obligations depend on the applicable jurisdiction and circumstances. Vendors do, however, need to explain their capabilities accurately and distinguish technical functionality from legal determinations.
For example, a platform may provide an audit trail documenting changes to an AI-generated narrative. That is a potentially valuable capability, but whether the audit trail satisfies a particular agency's preservation obligations is a separate question. Similarly, a system may allow officers to certify that they reviewed a report without necessarily establishing that the agency's verification process meets every applicable legal or prosecutorial expectation.
The distinction between supporting compliance-related workflows and guaranteeing legal compliance needs to be clear in vendor messaging.
For companies serving agencies across multiple jurisdictions, this has implications for sales enablement. Sales teams need to understand which capabilities may be relevant to disclosure, verification, and records preservation, how to explain those capabilities without making unsupported legal assurances, and when an agency needs to consult its prosecutor or legal counsel.
The objective is to equip sales representatives to communicate accurately about the product while recognizing where agency-specific decisions begin.
What does the complete reporting workflow look like?
The Police1 article emphasizes that officers must be able to explain and defend the contents of reports prepared with AI assistance. AI-generated narratives can contain errors, including details that are inconsistent with source material. Officers therefore need to understand how the technology uses available information and how to identify and correct inaccuracies.
These concerns are relevant to what a product demonstration communicates about the complete reporting process. A demonstration may show how quickly the system generates a narrative, how an officer reviews that narrative, how corrections are documented, and how the final report is verified. Different vendors will emphasize different aspects depending on their products, audiences, and the stage of evaluation.
The important consideration is whether the demonstration gives the buyer enough information to understand the workflow and the responsibilities that remain with the officer.
Consider a hypothetical example in which an AI-generated report describes an individual as resisting an officer's commands, but the available recording does not support that characterization.
An agency evaluating the technology would need to understand how an officer identifies the discrepancy, whether the officer can locate the source material associated with the statement, how the system handles the correction, and what record is preserved of the original language and subsequent revision.
The answers depend on the platform's capabilities and configuration. Explaining those capabilities helps prospective customers evaluate how the technology fits into existing reporting and supervisory processes.
It also highlights the distinction between product functionality and how that functionality is communicated. A company may have developed substantial verification and auditing capabilities that are thoroughly covered during technical evaluations. Those same capabilities may receive less attention in website copy, product descriptions, or early sales conversations.
Buyers encounter a company's messaging long before they reach a detailed product evaluation. The information available at each stage shapes their understanding of what the product does and whether it warrants further consideration.
Evidentiary capabilities have implications for messaging
Audit logs, version histories, records retention, data export, and access controls may appear to be technical features, but the Police1 article illustrates their operational significance.
Police executives are being encouraged to ask whether their agencies can preserve original drafts, retrieve relevant records, comply with litigation holds, and produce information years after a report was created. They also need to understand who can access agency data, how vendor subcontractors handle information, whether system settings can change, and what happens to records when a contract ends.
These concerns extend beyond the officer using the product. Records personnel may need to retrieve historical material, prosecutors may need information about how a report was prepared, and legal advisors may need to evaluate retention and disclosure obligations. Command staff need to understand whether system configurations align with agency policies.
For vendors, the challenge is determining how to communicate relevant capabilities to different stakeholders without overwhelming them with technical detail. A patrol supervisor evaluating the officer's review process may need different information than a records manager assessing retention and retrieval. A police chief considering organizational risk may need to understand how those capabilities work together.
A statement that a platform maintains an audit trail may be appropriate in introductory marketing. During a more detailed evaluation, however, the agency may need to know precisely what events are recorded, whether original drafts are preserved, who controls retention settings, and how records can be exported.
The company's website, sales materials, product demonstrations, and technical documentation serve different purposes. They do not need to contain identical levels of detail, but they should present a consistent and accurate account of the product's capabilities.
That consistency is particularly important when buyers must evaluate technology across operational, legal, and administrative responsibilities.
Pilot programs can help substantiate the value proposition
Pilot programs offer vendors an opportunity to demonstrate how their products perform in actual agency workflows. But as the Police1 article emphasizes, testing AI-assisted report writing on live cases carries legal and operational consequences. Agencies must consider policies, training, supervision, records retention, and permitted uses before deployment, even when that deployment is temporary.
These considerations should inform how pilots are designed and evaluated. A pilot focused exclusively on how quickly the software generates a narrative may provide useful information about drafting speed, but it won't necessarily establish whether the technology improves the complete reporting process.
For vendors, the opportunity is to work with agencies to identify meaningful evaluation criteria. If efficiency is a primary selling point, those criteria might include total report completion time, including review and correction. If verification capabilities are central to the product's value, the evaluation might examine how officers identify discrepancies, access source material, and document revisions. Agencies may also need to assess records retrieval, retention, and export capabilities.
Officer and supervisor experiences, the frequency and types of corrections, and any additional work created for records personnel may provide further insight into how the technology performs.
The resulting evidence can strengthen vendor messaging, provided the findings are presented within the limits of the pilot's design. Results from one agency or deployment should not be generalized to other settings without supporting evidence.
A well-designed pilot can help an agency make a more informed adoption decision while giving the vendor a stronger basis for explaining what its technology accomplishes under actual operating conditions.
Does the value proposition reflect the buyer's decision?
The broader lesson from the IACP presentation extends beyond AI report writing. Public safety technology companies often communicate value through operational improvements such as speed, efficiency, automation, and reduced administrative work. Those benefits are relevant, but they may represent only part of what agency leaders need to evaluate.
For AI-assisted report writing, the purchasing decision involves questions about officer accountability, evidentiary integrity, records preservation, prosecutorial expectations, and implementation responsibilities. Some vendors may already address these issues extensively. Others may find opportunities to make existing capabilities more visible or improve how their messaging supports different stages of the buying process.
The useful exercise is to compare what prospective customers are being encouraged to evaluate with what the company communicates. That includes whether the value proposition reflects the full reporting workflow, whether efficiency claims are supported by clearly defined measurements, and whether buyers can understand the product's verification and preservation capabilities.
The initial draft may take seconds. The agency's responsibility for that report can last years.
For vendors developing and marketing AI-assisted report-writing tools, the opportunity is to ensure that the story they tell about their products reflects the full scope of the decisions their customers must make.
Does your messaging reflect what your buyers need to know?
Your product may already address the operational, technical, and compliance-related concerns your customers are evaluating. But are those capabilities clearly communicated across your website, sales materials, and product demonstrations?
Word Sleuth Agency helps public safety and justice technology companies identify gaps between what their products deliver and how they communicate their value.
Start with a Marketing Second Opinion