Heber City AI Police Report Generates Fictional Content from Background Audio (2025)

Incident Summary

In late 2025, the Heber City Police Department in Utah began using AI-assisted report writing tools. During use, an AI system generated an official police report containing the fabricated statement that an officer had “turned into a frog.”^[1]

Investigation revealed that the AI system had captured background audio from the Disney animated film “The Princess and the Frog,” which was playing in the vicinity during the recording. The system incorporated dialogue and narrative elements from the film into the police report as though they were factual observations.^[2]

The error was identified during human review of the generated report and corrected before it entered the official record. The incident received media coverage in January 2026.^[3]

Key Facts

• Context: AI-assisted police report writing deployed in Heber City, Utah (December 2025)
• Error: AI system incorporated dialogue from a Disney film into an official police report
• Cause: Background audio from “The Princess and the Frog” was captured and processed as relevant input
• Detection: Error caught during human review of the generated report
• Impact: Report corrected; no erroneous content entered the official record

Threat Patterns Involved

Primary: Overreliance & Automation Bias — The incident demonstrates the risks of deploying AI transcription and report generation tools in uncontrolled audio environments without adequate filtering or validation mechanisms.

Significance

While the immediate harm was limited due to the error being caught in review, this incident raises substantive concerns about AI-assisted documentation in law enforcement:

1. Environmental audio contamination — AI transcription systems operating in real-world environments must contend with ambient noise, background media, and overlapping audio sources. This incident demonstrates that current systems may lack the ability to distinguish relevant speech from irrelevant background audio.
2. Downstream consequences of undetected errors — Had this error not been caught during review, a factually absurd statement would have been entered into an official police report. In cases involving less obviously fictional content, subtle AI-generated errors could go undetected and affect legal proceedings.
3. Automation bias in high-stakes contexts — Police reports serve as foundational documents in the criminal justice system. The introduction of AI-generated content into these documents necessitates robust review processes to prevent automation bias from undermining report accuracy.
4. Pilot program value — The fact that this error was surfaced during a controlled pilot, rather than in full deployment, underscores the importance of phased rollouts with human oversight for AI tools in high-stakes institutional settings.