AINEWS
At the May 2026 IACP (International Association of Chiefs of Police) conference, vendors sold facial recognition, license plate readers, and decision assist tools. Federal standards are missing.
For two decades, the live wire of American policing was the camera. Body cams, dashboard cams, automated license-plate readers, and gunshot detectors all poured data into real-time crime centers that, until recently, still relied on human analysts to decide what to do with it. The new product at the largest US police trade show is no longer the camera. It is the decision.
At the May 2026 International Association of Chiefs of Police (IACP) Technology Conference in Fort Worth, Texas, vendors sold that future to thousands of law enforcement buyers: facial-recognition cameras that flag a match before an officer makes contact; automated license-plate readers that score cars against hot lists; report-writing tools that draft a narrative from body-cam audio; chatbots that handle non-emergency 911 calls; and AI "decision-assist" platforms that suggest where to send a car or whom to stop. The Verge's Webb Wright, who reported from the IACP floor, was denied press entry and reconstructed the showroom from nearby hotels via attendee interviews.
The selling pitch borrows from air-traffic control. ForceMetrics, which sells an AI-powered decision-assist platform called Velocity RTCC, told The Verge the goal is a "centralized digital brain" for policing. The company's founder is careful with language: he avoids the "p word" (predictive policing) because the brand is "burned." Real-time crime centers themselves are not new; the New York Police Department adopted the model more than twenty years ago, and by 2019 was collecting roughly two years' worth of body-camera footage every week. What is new is that the analyst in the chair is being replaced by software.
Brookhaven, Georgia, Captain Abrem Ayana was at the IACP show. On the record, he called "a lot of it" sales gimmicks that do not deliver on the promise. His pushback is the kind of signal that should embarrass the sales force, and is instead a feature of the marketplace, because the buyer is also the regulator and there are no comprehensive federal standards or industry rules to fall back on. As The Verge reports, departments have "no choice but to take companies' word."
The pattern is not new. CompStat, the New York Police Department's statistics-driven deployment system, was shown in 2012 to have been manipulated by supervisors gaming the numbers. PredPol, the predictive-policing startup, was dismantled in 2020 after independent reviews found it reinforced the bias of its inputs. The Verge cites a 2012 New York Times investigation and a 2020 MIT Technology Review piece for those failures. Both earlier tools were sold as neutral math. The current wave is sold the same way, but with a sharper knife: the old tools still left a human at the helm on consequential calls. The new ones move the call itself into the model.
The market is large and moving fast. Axon Enterprise reported Q1 2026 revenue of $807 million, up 34 percent year over year, with Software and Services revenue of $355 million, up 35 percent. Inside that, the company said "AI products" revenue was up more than 700 percent year over year. Counter-drone revenue was up more than 300 percent; net revenue retention was 125 percent and annualized recurring revenue reached $1.5 billion, up 35 percent. The unit-of-sale shift shows up in the product names: Axon Vision fuses data from cameras, ALPRs, and gunshot detectors; Draft One and Brief One use generative AI to write police reports from body-cam audio; Dedrone handles airspace; and Fusus, the real-time-operations platform Axon acquired in February 2024, is the substrate underneath the rest. Axon raised its full-year 2026 revenue outlook to 30 to 32 percent growth, with operating cash flow above $600 million and free cash flow around $450 million.
The federal rule that would make a department reckless to ignore Ayana's warning has not been written.