Can police in California create police reports using an AI Engine?
In California criminal defense attorneys, car accident lawyers and sometimes divorce lawyers rely upon police reports for use in motions and in court. Police reports are standard forms (for each department) and are generated from notes taken in the field and the officer’s memories. Lately an increasing number of officers are using AI and dictation into the AI engine to create police reports.
In October 2025, California Governor Gavin Newsom signed Senate Bill 524 (SB 524), authored by Senator Jesse Arreguín and sponsored by the California Public Defenders Association. This law, effective January 1, 2026, introduces transparency requirements for law enforcement agencies using generative artificial intelligence (AI) to draft or assist in writing official police reports. It makes California the second state (after Utah) to regulate this emerging practice.
Key Provisions:
Mandatory Disclosure: Any police report created fully or in part using AI must include a clear written statement, such as "This report was written either fully or in part using artificial intelligence." This disclosure is required on every page or prominently within the report text.
Audit Trail and Retention: Agencies must preserve the initial AI-generated draft for as long as the final report is retained. They must also maintain records identifying the officer responsible, the AI tool/version used, and links to supporting materials like body-worn camera audio/video footage.
Officer Accountability: Officers must review, edit if needed, and certify the accuracy of AI-assisted reports with their signature.
Data Protections: AI vendors (e.g., companies like Axon, maker of the popular "Draft One" tool) are prohibited from selling, sharing, or otherwise using data provided by law enforcement for AI processing, except as necessary for the service.
Background and Purpose:
The law addresses growing concerns over tools like Axon's Draft One, which transcribe body-camera audio and generate narrative drafts to save officers time. Critics, including the Electronic Frontier Foundation (EFF) and public defenders, worry about potential AI errors, biases (from training on historical reports), hallucinations, or reduced accountability in criminal proceedings—where reports heavily influence charging decisions, trials, and outcomes.
Proponents emphasize that the law promotes due process by ensuring judges, prosecutors, defense attorneys, juries, and the public know when AI contributed to a report's narrative. It does not prohibit AI use in report writing but adds safeguards for transparency and auditability.
Initial opposition from police associations (citing administrative burdens) led to amendments narrowing the scope, after which support grew. Advocates view SB 524 as a foundational step, with calls for stronger future regulations, potentially including broader restrictions on AI in policing.
Discovery Demand Ideas
In most cases the process of creating a police report is not a critical issue. In cases where you suspect the report has been altered, materially changed or is missing data you should make very specific PC 1054.1 discovery requests. You should ask for:
1. Production of the Police Report in its Native Format with all meta data preserved.
2. All e-mails and texts relating to the report e.g. where one officer asks the officer by text do you remember ........................ or Did the suspect say ....................
3. The identity of anyone who reviewed the report before it was final and get copies of what that person(s) reviewed.
4. For AI engines that keep track of your inquiries request copies of all inquiries and responses to the AI engine.
Technology is an increasingly important part of criminal defense and our attorneys are expert in technology. Daniel Horowitz was a beta tester for multiple attorney document management software solutions and a trainer in their use. He is also a Criminal Defense Specialist authorized to use that phrase by the Board of Legal Specialization for the State Bar of California. Call Daniel Horowitz at (925) 283-1863
This is part of California's broader 2025 push on AI ethics, alongside laws on deepfakes, safety testing for large models, and other uses. No outright ban on AI-generated police reports exists in California as of late 2025.
