AI Misfire: Teen Handcuffed After AI Mistakes Doritos for Gun

AI error leads police to handcuff teen after mistaking Doritos for a gun, raising new concerns over ethics in school surveillance systems.

A 16-year-old student in Baltimore, Maryland, was handcuffed by police after an artificial intelligence (AI) system wrongly identified a bag of chips as a firearm.

The incident has reignited debate over the accuracy, reliability, and ethics of AI-based weapon detection systems in U.S. schools.

“Police showed up, like eight cop cars, and then they all came out with guns pointed at me talking getting on the ground,” said student Taki Allen, in an interview with local media outlet WMAR-2 News.

The event underscores the risks of deploying untested or overconfident AI surveillance tools in sensitive public spaces.

According to the Baltimore County Police Department, officers “responded appropriately and proportionally based on the information provided at the time.”

However, later review revealed the alert was a false alarm — the system had confused Allen’s chip bag for a firearm.

BBC reported that the AI alert, provided by Omnilert’s technology, was reportedly sent to human reviewers who found no threat.

Yet, the school principal failed to see the “no threat” update and contacted the school resource officer, who in turn called local law enforcement.

The miscommunication led to the arrival of armed officers on school grounds, escalating a non-event into a traumatic experience for the student.

In a statement to BBC News, Omnilert said it “regrets this incident occurred and wishes to convey our concern to the student and the wider community affected.”

The company emphasized that its system “operated as designed” and that its human verification process worked correctly — the failure, it said, came later in procedural handoff.

While Omnilert defends its technology, the company admits that “real-world gun detection is messy.”

AI models rely on training data that may not encompass every lighting condition, object shape, or color variation.

In this case, the system’s visual model apparently could not distinguish the reflective surface of a chip bag from a firearm.

The misidentification highlights a growing problem with AI in safety and law enforcement — false positives that can lead to dangerous or traumatic consequences in the real-world.

Cybersecurity governance now extends beyond data privacy and system security — it must also ensure ethical AI deployment.

This includes auditing algorithms for bias, testing for real-world accuracy, and establishing transparent escalation procedures.

Without proper oversight, the rapid rollout of AI surveillance tools could amplify human error rather than reduce risk.

AI ethicists argue that systems intended to protect should undergo the same level of scrutiny as cybersecurity defenses.

To prevent similar incidents, school districts and organizations adopting AI detection tools should apply a layered approach that balances safety with ethical responsibility:

Together, these measures help ensure AI-driven security systems operate responsibly, minimizing harm while maintaining trust.

As AI technologies expand into policing, hiring, and education, their errors can carry disproportionate consequences. Baltimore’s chip incident illustrates how a system meant to prevent violence can instead inflict harm through misinterpretation and procedural failure.

The rapid adoption of AI in schools and public safety sectors demands stronger regulatory frameworks, standardized accuracy testing, and third-party auditing.

Mistakes like this highlight why ethical oversight is no longer optional — it is a fundamental requirement of safe AI deployment.

The Baltimore incident serves as a cautionary tale for all organizations integrating AI into decision-making and security processes.

As AI systems grow more autonomous, human accountability must keep pace. The future of cybersecurity and public safety lies not just in advanced algorithms — but in ensuring those algorithms are fair, transparent, and trustworthy.

As artificial intelligence becomes harder to distinguish from reality, investing in reliable deepfake detection tools is becoming essential for digital safety.

Ken Underhill is an award-winning cybersecurity professional, bestselling author, and seasoned IT professional. He holds a graduate degree in cybersecurity and information assurance from Western Governors University and brings years of hands-on experience to the field.

The new Android Trojan Herodotus mimics human behavior to evade modern anti-fraud systems.

A critical Magento flaw, SessionReaper (CVE-2025-54236), is exploited in the wild. Learn how to patch and protect your e-commerce systems.

A new Ubuntu kernel flaw lets local attackers gain root access through patch inconsistencies.

BlueNoroff uses AI-driven attacks in GhostCall and GhostHire to target global crypto and tech executives.