Finding Transparency Through AI Labels

Artificial intelligence has dominated the conversation around technology and communications over the last few years, and it doesn’t appear to be going anywhere anytime soon. One big question has been how law enforcement agencies better articulate what AI does for their staff members and community stakeholders.

This summer, Motorola Solutions announced the introduction of ‘AI labels’ into its ecosystem of products to help agencies better understand how AI is used in its software and hardware solutions. The initiative takes the idea of a common food label, listing the “ingredients” or types of AI used and how data is treated and stored.

“It is our unwavering conviction that technology — including AI — is the bedrock for safety and security, and it must be deployed with purpose and transparency to fulfill its promise as a force for good,” Mahesh Saptharishi, Executive Vice President and Chief Technology Officer at Motorola Solutions, said in a statement. “Nutrition labels help describe AI’s use in protecting neighborhoods and nations, and we are proud to take a lead role in bringing greater transparency to AI innovation.”

During APCO 2025 in Baltimore, which was held just days following the announcement, Jim Wolfinbarger, retired Chief of the Colorado State Patrol who now serves as Vice President of Software Sales at Motorola Solutions, spoke to OFFICER Magazine about the AI labels initiative and how it seeks to bring clarity and transparency to a topic that is often clouded in mystery.

“We see all of these things that are ‘AI enabled,’ but what does that really mean?,” he says. “How are you using it? Who owns the data? Is it using a third party or first party model?”

Bringing clarity to AI

While the labels aren’t literally on the product boxes — as they were for an APCO display — Wolfinbarger says they will be an important element of Motorola Solutions’ sales and education process. “We’re using the metaphor for cereal boxes, or any sort of food, to better understand what’s actually inside of it. There’s a lot of times where food may be misleading in regard to how nutritious or beneficial it actually is or isn’t. A good way to kind of get to the source of the truth is to be able to consult a label.” He added that customers can view the labels on the product pages on Motorola Solutions’ website. “You can view the different labels for both the hardware and software. Everything that Motorola Solutions comes out with will have an associated AI label with it.”

He says that the goal of the AI labels is to create transparency. “They provide a base level of understanding between us as a vendor and our customers to have honest and transparent conversations about what artificial intelligence is; how it affects the technology they’re introducing into their community and department; and to be able to have very informed conversations about what it is.”

Each label explains the type of AI used, who owns the data processed, human controls and the purpose behind the product’s specific application of AI. The AI nutrition labels are an initiative of the Motorola Solutions Technology Advisory Committee (MTAC), a group that serves as the company’s ‘technical conscience’ and guides it on ethics, limitations and implications of specific technologies.

An education tool

While Motorola Solutions plans to use the labels to educate potential customers, Wolfinbarger says law enforcement agencies can also make use of them to educate lawmakers, community members and their own officers.

“In this day and age, with so much rapid change in technology, I think that having clear, concise ways of being able to move through discourse and conversations are really important,” he says. “As we update anything inside our programs, we’ll update the corresponding AI label so that there’s always this transparency between Motorola Solutions and its customers.”

He added that when agencies are engaging with community privacy advocacy groups like the ACLU on AI and the technology that is being procured, the labels can be useful. “It’s super easy to pull that out and show exactly what you’re using and how you’re using it.” He says the labels also can be a tool when speaking with members of the community, lawmakers and employee unions.

Varying ingredients

As ingredients in different food products vary, the same goes for AI products. “With transparency being the primary purpose behind the labels, Wolfinbarger stressed the importance of clarity in how information is shared with customers. “First and foremost, there is the description about what the product actually does, and then some understanding around what are the model purposes of AI application within the context of that description, and then some specifics about the component elements of it. Is the model trained with customer data?”

In the case of CommandCentral DEMS, which was used as an example, the answer is “No,” which means the AI-enabled software uses internal data from Motorola Solutions and select non-customer datasets. CommandCentral DEMS does use Generative AI for object detection, redaction, summarization and other purposes outlined on its label.

In other products, like the company’s speaker microphone and video camera combo SVX device, there is no Generative AI used, but it does use Edge AI for noise suppression. “Through our AI labels initiative, you get a strong sense of the AI used within Motorola Solutions’ products,” he says.

Another thing that Wolfinbarger says law enforcement agencies need to keep in mind when it comes to artificial intelligence is the evolving meaning of the terms being used, which will likely change over time. “When it says ‘Audio/written summarizations,’ that’s the idea of being able to take all the radio traffic and all the 911 traffic and cross-reference that speech to text and then to organize that text in such a way that we can identify keywords, and then provide all of that content from multiple sources to assist in building out a narrative that’s accurate and comprehensive. Some of the early iterations of AI-generated report writing typically relied on a single source,” he says. “That was great in the early days, but the concern always is that a single source of data can be dangerous. Bringing together multiple sources, where you are collecting from 911, radio traffic, written reports and CAD (Computer Aided Dispatch), can help provide a report that is more complete and more accurate.”