Share with your CISO
A gold rush of AI vendors is selling police departments facial recognition, automated license plate readers, AI-written reports, and real-time crime centers (RTCCs) that aggregate surveillance data into algorithmic decision-making. Axon Enterprise, whose AI product revenue grew 700 percent year over year, is competing with Motorola Solutions and a wave of startups to own the entire law enforcement technology stack. The market is large, the contracts are long, the regulatory oversight is essentially nonexistent, and the failure modes from earlier predictive policing programs have not been resolved, only rebranded.
What this means for your business
If your organization sells to, partners with, or operates infrastructure for law enforcement agencies, this procurement environment is now your risk surface. Axon’s AI Era Plan, a flat-fee subscription bundling current and future AI tools, is the vendor playbook writ large: lock in the department, then expand the product footprint before governance catches up. Any vendor whose data, cloud, or identity systems touch a police RTCC inherits the legal and reputational exposure when that system produces a wrongful outcome. The question is whether your contracts price that liability correctly.
The deeper structural problem is what might be called bias laundering, where historical data infected by decades of racially uneven enforcement gets fed into new AI systems that present their outputs as objective. PredPol failed this way. CompStat failed this way. The vendors at the IACP Technology Conference are now selling systems that ingest body camera footage, 911 logs, parole records, and license plate reader data into RTCCs, and asserting that the failures of the past were a data-volume problem rather than a data-quality problem. That framing is convenient for vendors whose business depends on selling more data-collection hardware alongside the analytics layer.
Axon’s Draft One episode is the clearest warning for anyone building AI into a legal or compliance workflow. The system originally preserved no copy of its AI-generated output after submission, by design, eliminating the audit trail that courts and opposing counsel would need. Axon updated that behavior only after regulatory pressure accumulated. Organizations deploying AI in any context where outputs become legal documents, including HR decisions, financial disclosures, or contractual records, should treat that as a live falsification test: if your vendor has designed away the audit log, you do not have human oversight, you have human credit-taking for machine output.
Concept deep-dive: Real-Time Crime Center (RTCC)
An RTCC is a centralized platform that pulls live data streams from surveillance cameras, 911 dispatch, license plate readers, and criminal records into a single operational dashboard, functioning like an air traffic control system for patrol decisions. The concept originated at the NYPD over 20 years ago with human analysts doing the aggregation. Modern RTCCs replace those analysts with AI, compressing the time between data ingestion and officer instruction. The business risk is that speed and opacity arrive together, with no mechanism to audit why a specific alert was generated.
Based on reporting from Computer cops: Inside the big business of selling AI to the police, originally published 2026-07-16 07:00:00.
