Key Takeaways
- Several Michigan police departments are piloting AI‑driven report‑writing tools, most notably Axon’s Draft One, to reduce the time officers spend on paperwork.
- Proponents argue the technology frees officers for community work, while critics worry it erodes essential writing skills and may compromise report accuracy.
- Studies from the University of South Carolina indicate Draft One speeds up initial drafting but does not save overall time because officers must still verify and correct AI‑generated text.
- Accuracy concerns have led some prosecutors (e.g., in King County, WA and statewide Connecticut) to reject or pause AI‑assisted reports; Wayne County Prosecutor Kym Worthy is drafting county‑wide guidance urging caution.
- Departments vary in policy: some allow AI for all incident types, others restrict it to non‑violent offenses, and many require officer review, supervisory approval, and explicit labeling of AI‑assisted sections.
- Officers are trained to spot deliberately inserted “Easter egg” errors (e.g., a suspect’s alibi involving a time loop) as a safeguard to ensure they remain actively engaged in the report‑creation process.
Adoption of AI Report‑Writing Tools in Michigan
A small but growing number of Michigan police departments have begun experimenting with artificial intelligence to help officers compose police reports. The technology is marketed as a time‑saving solution for agencies stretched thin by staffing shortages and rising call volumes. Early adopters such as Dearborn Heights and the Warren Police Department have integrated Axon’s Draft One into their workflow, hoping to shift officers from desk work back onto the streets. As Captain Jake Hatten of Dearborn Heights explained, “It’s still the officer’s report… It’s not AI’s report, it’s the officer’s report.”
How Draft One Works
Draft One transcribes audio captured by officers’ body‑worn cameras and then generates an AI‑written summary of the incident. Officers are expected to review the draft, edit out any inaccuracies, and add missing details before finalizing the report. To ensure active participation, the software inserts deliberate mistakes—referred to internally as “Easter eggs”—such as the claim that “the primary suspect was absolved by presenting a convincing alibi involving a time loop.” Officers must locate and remove these fabricated elements, a process designed to keep them engaged with the content.
Perceived Benefits for Police Agencies
Supporters of the tool highlight its potential to reclaim officer time. Warren Police Commissioner Eric Hawkins noted, “If I can cut down on the time that an officer writes a report… I can get them back into the community—to me, that’s a bonus.” In a demonstration observed by 7 News Detroit, Hawkins added that the technology “makes it easier for them to do the work,” but stressed that officers “they actually have to do the work and they have to be accountable for what’s being done.” The belief is that by reducing paperwork, officers can spend more time patrolling, responding to calls, and engaging with the public.
Research Findings on Time Savings
Contrary to the marketing promise of efficiency, academic research suggests the time‑saving benefits may be illusory. A study conducted by the University of South Carolina found that while Draft One produces a draft quickly, officers still needed to spend considerable time verifying the AI‑generated text and correcting errors. Lead researcher Ian Adams, an associate professor, stated, “It didn’t actually save officers any time on their reports.” The study concluded that any initial speed gain was offset by the subsequent verification workload.
Impact on Report Accuracy
Beyond time concerns, the same USC research team examined the substantive quality of AI‑assisted reports versus those written solely by officers. They discovered that AI‑generated narratives tended to use more sophisticated vocabulary—written at a higher grade level—but were rated “substantively and significantly lower on accuracy” than officer‑authored reports. Commissioner Hawkins reacted to this finding with concern, admitting, “I did not know about that… but just hearing about that, yeah of course that causes me concern.” The results prompted him to advocate for a cautious, data‑driven pilot program before any full‑scale rollout.
Officer Skill Variations
Early feedback from the Warren pilot revealed a split in user experience. Officers who already possessed strong report‑writing abilities saw little advantage from the AI tool, while those with weaker writing skills reported substantial improvements. Hawkins described the outcomes as “mixed,” noting, “We’re seeing some officers who are very, very skilled in writing reports that aren’t seeing much benefit from this… We’re seeing some officers who may have lesser skills to writing reports who are seeing huge benefits.” This disparity suggests the technology may serve as a crutch for less experienced writers rather than a universal efficiency booster.
Varied Departmental Policies
Agencies differ widely in how they permit AI use. In Warren and Dearborn Heights, officers may employ Draft One for any type of incident, ranging from minor traffic stops to serious felonies. Conversely, the Kent County Sheriff’s Office restricts the tool to non‑violent offenses, citing prosecutorial comfort with the resulting documents. Undersheriff Bryan Muir observed that prosecutors “have actually commented they like to see the AI reports… They typically are more thorough, more comprehensive than some of the typed reports our officers are providing.”
Prosecutorial Skepticism and Restrictions
Other jurisdictions have taken a far more restrictive stance. In King County, Washington, prosecutors refuse to accept cases where reports were generated with AI assistance. Across the state of Connecticut, all AI‑assisted reports are currently on pause pending further review of accuracy concerns. In Wayne County, Prosecutor Kym Worthy is in the process of drafting a county‑wide policy that will mandate thorough verification of any AI‑generated content. In a recent letter to police chiefs, she wrote, “under no circumstances should AI be used without thoroughly checking and double-checking its authenticity,” warning that reliance on unchecked AI could jeopardize prosecutions.
Safeguards and Best Practices
Prosecutors and law‑enforcement leaders alike emphasize the need for concrete safeguards when using AI for report drafting. Recommended practices include: clearly labeling reports as AI‑assisted, requiring officers to perform meaningful edits and reviews, supervisory sign‑off, and routine testing with intentionally obvious errors (the “Easter eggs”). The Macomb County prosecutor’s office echoed this approach, stating they are “not imposing a one‑size‑fits‑all policy” but are encouraging agencies to consult before implementation to discuss benefits, challenges, and appropriate controls.
Local Hesitation and Skill Preservation
Many Michigan departments remain wary of rapid adoption. Southfield Police Chief Elvin Barren articulated a philosophical objection, saying, “We’re in the business of producing future leaders… I want competent leaders who are comfortable with writing, comfortable with expressing themselves and not—over a period of time—allowing a system or AI to replace those skills.” His stance reflects a broader concern that overreliance on AI could erode fundamental communication abilities that are vital for testimony, community interaction, and leadership development.
Conclusion
The integration of AI‑driven report‑writing tools like Draft One presents a nuanced picture for Michigan law enforcement. While the technology offers a tempting shortcut for busy departments, empirical evidence shows it does not eliminate the workload associated with report creation and may introduce accuracy issues. Prosecutors are urging caution, drafting policies that demand verification and accountability, and some jurisdictions have outright banned AI‑assisted reports. Departments that continue to experiment are doing so with varied restrictions, officer training, and oversight mechanisms. Ultimately, the conversation centers on balancing potential efficiency gains against the preservation of essential writing skills, evidentiary reliability, and the ethical responsibility officers bear for the documents that underpin criminal justice proceedings.
https://www.wxyz.com/news/local-news/investigations/michigan-officers-turn-to-ai-for-police-reports-wayne-co-prosecutor-warns-be-cautious