CrimeRadar AI App Sends False Crime Alerts Across U.S. Communities (2025)

Incident Summary

In December 2025, the CrimeRadar app — an AI-powered tool developed by Scoopz Inc. with over 2 million users — sent waves of false notifications about shootings and violent crimes to residents across multiple U.S. cities.^[1]

The app uses AI to monitor publicly accessible police radio transmissions, convert live audio to text, and generate real-time crime alerts. However, its speech-to-text system repeatedly misinterpreted routine radio chatter with dangerous consequences. In Streetsboro, Ohio, a kindergartener pulling a fire alarm became “firearms discharged at elementary school,” nearly triggering a panic. In Bend, Oregon, a “Shop With the Cop” charity event was reported as an officer being shot. In Columbia, Missouri, a false alert claimed a 17-year-old had been shot downtown, drawing from dispatch audio from multiple unrelated counties.^[2]

A BBC Verify investigation documented the pattern, finding evidence of the app “repeatedly sending out misleading, inaccurate, or false crime alerts” across communities from Florida to Oregon. Columbia Police Chief Jill Schlude issued a public statement warning that AI-powered crime apps “often rely on incomplete data, lack context, and can easily spread misinformation.” CrimeRadar apologized for the distress caused and promised upgraded audio processing and community correction features.^[3]

Key Facts

• App: CrimeRadar by Scoopz Inc. (Mountain View, CA); over 2 million users; ~700,000 downloads in one recent month
• Mechanism: AI speech-to-text converts police radio to alerts; system misinterprets routine chatter as violent crime reports
• Key false alerts: “Firearms discharged at elementary school” (fire alarm pull), “officer shot” (charity event), false teen shooting (audio from wrong counties)
• Cities affected: Streetsboro OH, Columbia MO, Bend OR, Boulder County CO, Marion County OH, plus reports from Florida, Tennessee, and other states
• Police response: Multiple departments issued public warnings; Columbia Police Chief warned about AI crime apps spreading misinformation
• Company response: Apology issued following BBC Verify investigation; promised upgraded audio processing and community correction features

Threat Patterns Involved

Primary: Misinformation and Hallucinated Content — AI speech-to-text systems misinterpreted police radio transmissions, generating false reports of violent crimes that were pushed to millions of users without verification.

Secondary: Overreliance and Automation Bias — The system published AI-generated alerts to users without human review or contextual verification, placing automated output directly into a public safety information channel.

Significance

1. Public safety information integrity — False alerts about school shootings, violent crimes, and officer injuries in a public safety context carry particularly high stakes, as they can trigger panic, alter behavior, and erode trust in legitimate emergency communications
2. Speech-to-text limitations in safety-critical applications — The specific failure modes (fire alarm → firearm, “Shop With the Cop” → officer shot) demonstrate that current speech-to-text AI lacks sufficient contextual understanding for safety-critical public alert systems
3. Scale amplification — With over 2 million users, each false alert reached a substantial audience, transforming individual transcription errors into community-wide misinformation events
4. Missing verification layer — The system pushed AI-generated alerts directly to users without human review, contextual verification, or confirmation from official sources, representing a design choice that prioritized speed over accuracy in a domain where accuracy is essential