A 16-year-old at Kenwood High in Baltimore County was detained at gunpoint after an AI security system flagged his bag of Doritos as a firearm on 20 October 2025.
📌 Key Takeaways
- False Alert: AI misread a crumpled chip bag as a gun, prompting an armed police response.
- Bodycam Released: Footage shows officers realizing the “weapon” was only chips.
- Vendor Identified: Local outlets name Omnilert as the system in use.
- Policy Review: County leaders and police are reviewing procedures after the incident.
What Happened Outside Kenwood High
Police converged on the student after an AI camera alert. He was ordered to the ground and cuffed. No weapon was found once officers searched the scene and questioned him.
Body-worn video released on 26 October shows officers second-guessing the alert before confirming it was a bag of chips, not a gun.
“It was like eight cop cars… with guns, saying get on the ground.” — Taki Allen, Student
How The AI Flagged A Snack As A Threat
Local coverage identifies the vendor as Omnilert, which analyzes video feeds for firearm-like shapes. In this case, a shiny, folded bag and hand posture were interpreted as a handgun.
Omnilert’s marketing materials emphasize high accuracy and low false positives, typically with human verification. The Kenwood incident shows how a single misread can escalate quickly.
“The cost to do animations is in the hundreds of thousands — this set-up is orders of magnitude smaller.”
Bodycam, Timeline, And The Immediate Aftermath
Officers responded around 7:20 p.m., detained the student, then cleared the scene once the error was discovered. The bag of chips was recovered nearby.
County leaders requested a formal review of the system and procedures. Police emphasized they acted on information available from the AI alert.
“Thank God it was not worse.” — Baltimore County Council Leadership
Accuracy Questions Beyond One Vendor
Independent reporting shows other school scanners can generate heavy noise. One Illinois district logged 3,248 alerts in a year with only 3 confirmed contraband items, a >99% false-positive profile. That system was Evolv, a different vendor.
Separately, the FTC sanctioned Evolv in 2024 over deceptive accuracy claims, underlining wider scrutiny of AI screening promises. Again, Evolv is not the Kenwood vendor.
What Baltimore County Is Reviewing Now
Here’s a quick bridge into the moving pieces being examined by officials and school leaders, based on public statements and local reporting.
- Alert Workflow: How AI alerts are verified before dispatch and what human checks exist.
- Officer Guidance: How responding units are briefed on AI uncertainty and image ambiguity.
- Threshold Tuning: Whether detection sensitivity or model retraining is needed at campuses.
- Student Impact: Counseling offers and communication protocols after traumatic false alerts.
“Humans will always check what the tool produces — that’s non-negotiable.”
Why This Incident Matters For AI In Schools
AI surveillance can expand coverage and speed response, yet misclassification risks are real. Clear verification steps and conservative escalation policies reduce harm from rare but high-impact mistakes.
The Kenwood case illustrates how perception errors cascade into police tactics. Transparent audits help determine where process, model tuning, or operator training should change.
Conclusion
Baltimore County is reassessing how AI alerts feed into real-world decisions. The goal is a safer campus, supported by technology that is verified, well-tuned, and paired with careful human judgment.
The Kenwood misfire is a cautionary data point. Reliability claims must match lived outcomes, especially when seconds and assumptions shape high-stakes responses.
📈 Latest AI News
27th October 2025
- Vibe Coding Goes Mainstream: Did Google Just Kill Every Paid Coding Tool
- Oreo Maker Mondelez Is Turning to Generative AI to Slash Marketing Costs
- Microsoft Launches AI companion Mico for Copilot
- Anthropic Strikes New Deal to Use Google’s AI Chips to Train Claude
- OpenAI Acquires Sky: An AI-Powered Interface for Mac Users
For the recent AI News, visit our site.
If you liked this article, be sure to follow us on X/Twitter and also LinkedIn for more exclusive content.
