California has just taken one of the boldest steps yet to bring transparency to artificial intelligence in law enforcement.
Governor Gavin Newsom has signed Senate Bill 524 (S.B. 524), a first-of-its-kind law requiring police departments to publicly disclose when AI tools are used to write or assist in writing official police reports.
The move, which takes effect on January 1, 2026, makes California only the second state in the U.S.—after Utah—to impose specific transparency rules on AI-written law enforcement documents.
Under the new law, every page of a police report that used AI must clearly display the statement:
“This report was written either fully or in part using artificial intelligence.”
It also forces departments to retain every draft of the report, including the original AI-generated version, so that courts, defense attorneys, and auditors can determine which sections were written by officers and which came from AI.
The bill—sponsored by Senator Jesse Arreguín (D–Oakland)—goes further, banning AI vendors from selling or sharing data submitted by police agencies. Vendors like Axon, which produces the popular “Draft One” AI report-writing tool, could now face serious compliance challenges.
A New Layer of Oversight
S.B. 524 is designed to shine light on a fast-growing but little-known corner of policing: AI-generated narratives. Some police departments have been quietly using AI systems to turn body-camera footage or officer audio notes into full reports.
Civil liberties advocates have long warned that this lack of transparency could lead to major due-process risks. “Police reports form the foundation of arrests and prosecutions,” the Electronic Frontier Foundation (EFF) wrote in support of the bill. “If AI is generating part of those reports, the public deserves to know.”
Critics point to systems like Axon’s Draft One, which does not retain edit logs showing which text came from the officer and which from AI. That design could now place agencies out of compliance with the new law unless vendors overhaul their systems or officers manually preserve every draft.
Why It Matters
Supporters say the measure is about more than paperwork—it’s about fairness, accountability, and public trust.
If an officer’s statement is partially AI-generated, defense attorneys could question whether the wording truly reflects what the officer observed, or whether the AI tool “filled in” missing context or phrased details in a biased way.
S.B. 524 creates a digital audit trail, requiring that each AI program used, the officer who initiated it, and any person who edited the text afterward all be logged. The law also mandates that any source material—such as video or audio used by the AI—be traceable.
A Blow to AI Vendors?
For police departments, the law creates a new logistical headache: they must now store every AI draft for as long as the final report exists. Vendors like Axon may need to redesign tools that currently overwrite earlier versions.
The EFF said it plans to monitor departments’ compliance closely and may file public-records requests to ensure AI use is being disclosed properly.
Meanwhile, police associations have voiced concern about the administrative burden of tracking every AI edit, but legal experts argue transparency is non-negotiable when technology begins shaping official statements that can determine someone’s freedom.
California Sets the Tone
With California joining Utah, other states are likely to follow. The state’s massive law enforcement system—over 500 agencies—means that vendors will almost certainly adapt products to comply nationwide.
The law also sends a global message: as AI becomes embedded in public institutions, lawmakers are moving from passive curiosity to direct regulation.
Legal analysts say the next step could include federal guidelines or even an outright ban on AI-generated law enforcement documents, if issues around accountability and evidence integrity persist.
What’s Next
S.B. 524 takes effect in January 2026, but departments will need to start preparing immediately—reviewing contracts, updating workflows, and ensuring their systems can retain version histories.
Civil-rights groups say this is only the beginning. “Transparency is step one,” said a senior digital-rights advocate at EFF. “Now we have to ask: should AI be writing police reports at all?”
For the public, it means that any future police report in California will have to clearly say whether AI had a hand in writing it—a small but significant shift toward openness in an era where algorithms increasingly shape real-world outcomes.
Key Takeaways
- California’s new law mandates AI-use disclosure in every police report.
- Police must retain original AI drafts and full edit histories.
- Vendors like Axon may need major redesigns to stay compliant.
- The law prohibits selling or sharing police data fed into AI.
- Takes effect January 1, 2026; EFF to monitor compliance.