A hot potato: Police departments are using a new tool that allows them to generate reports using generative AI. The software vendor that provides the tool claims its approach has solved some of the trickier problems associated with generative AI, such as hallucinations. However, the technology has yet to face scrutiny in the courts, meaning the debate over its use is far from over. Discussions are expected to revolve around issues of privacy, civil rights and justice.
Police departments, long accustomed to using technology in their workflows, have recently begun incorporating generative AI into their reporting processes. This shift follows the launch of Draft One, a new tool developed by police equipment provider Axon earlier this year.
Draft One uses Microsoft’s Azure OpenAI platform to transcribe audio from police body cameras and create draft reports. To ensure accuracy and objectivity, these reports are based strictly on audio transcripts and avoid any form of speculation or embellishment.
Officials must review and approve the reports, which are flagged to indicate an AI’s involvement in their creation. Although Axon uses the same underlying technology as ChatGPT, which is known to occasionally produce misleading information, the company claims to have fine-tuned the AI to prioritize factual accuracy and minimize hallucinations.
“We use the same underlying technology as ChatGPT, but we have access to more knobs and dials than an actual ChatGPT user,” Noah Spitzer-Williams, who manages Axon’s AI products, told The AP. Turning down the “creativity dial” helps the model stick to the facts so it “doesn’t exaggerate or hallucinate in the same way that you would if you were just using ChatGPT,” he said.
The selling points for Draft One are clear. It is marketed as a tool that will significantly reduce the amount of time officers spend on paperwork, and that can potentially reduce report writing time by 30 to 45 minutes per report. This efficiency, Axon says, allows officers to spend more time on community engagement and decision-making, which could improve de-escalation outcomes.
To be fair, officers who have used Draft One report that these AI-generated documents are not only time-efficient, but also accurate and well-structured. What’s more, the AI sometimes captures details that officers might miss.
Police departments across the country are using Draft One in different ways. In Oklahoma City, AI-generated reports are currently limited to minor incidents that do not involve arrests or violent crimes, following local prosecutors’ recommendation to proceed with caution.
In cities like Lafayette and Fort Collins, however, AI use is already more widespread and even covers large-scale projects. However, challenges remain, such as dealing with noisy environments.
Axon has not disclosed how many police departments currently use its technology. The company is not alone in this market, as startups like Policereports.ai and Truleo also offer comparable solutions. However, due to Axon’s established connections with law enforcement agencies that purchase its Tasers and body cameras, industry experts and police officials expect the use of AI-generated reports to become more widespread in the near future.
Despite its potential, the use of AI in report writing has raised concerns among legal scholars, prosecutors and citizen activists. They worry about AI’s tendency to “hallucinate” or produce false information, and the possibility of AI altering important documents in the criminal justice system. Other concerns include the embedding of societal biases in reports and the risk that officials will rely too heavily on AI, potentially leading to less thorough report writing.
“I fear that the automation and ease of use of the technology may cause police officers to be less careful in their writing,” says Andrew Ferguson, a law professor at American University who is working on what appears to be the first law journal article on the new technology.
