Baltimore Student Detained After AI Misidentifies Doritos Bag as Firearm

Post by : Raina Nasser

An incident in Baltimore County, Maryland has reignited international scrutiny of AI-driven surveillance after a teenage student was taken into custody when an automated system misclassified his bag of Doritos as a gun. The student, identified as Taki Allen, was outside Kenwood High School eating with friends when an AI alert precipitated a rapid police response — officers arrived with weapons drawn.

The alert originated from an AI monitoring platform developed by Omnilert, which reported what it assessed to be a weapon in Allen’s hand and notified school and law enforcement officials. Officers reached the scene quickly and ordered the student to lie on the ground. Allen later described the moment: “At first, I didn’t know what was happening until they started walking toward me with guns, shouting, ‘Get on the ground,’” adding that he had only been holding a Doritos bag.

Police reports indicate Allen was handcuffed and subjected to a search; no firearm was discovered. Subsequent review of surveillance footage showed the object flagged by the system was a crumpled orange chip packet. One officer noted the AI misread the way the bag was held, producing a false positive.

The episode has prompted immediate concern about the accuracy and deployment of artificial intelligence in school safety systems. Parents and students voiced alarm, demanding explanations for how such a misidentification could trigger an armed police intervention. “This could have gone horribly wrong,” said a parent gathered outside the school, stressing that human lives should not be put at risk by automated errors.

Kenwood High School’s administration issued an apology to Allen and to those who witnessed the event, acknowledging the distress it caused. The statement noted that counseling services have been made available to pupils affected by the episode.

Omnilert, the vendor of the detection software, issued a public expression of regret and confirmed it has opened an internal review of the visual recognition processes that produced the alert. The company said it is assessing ways to reduce erroneous identifications.

Experts argue the incident highlights broader questions about reliance on AI for security decisions in public spaces. Critics point out that current systems can lack context awareness and the nuanced judgement necessary for policing, increasing the risk of harmful outcomes when automated alerts are treated as definitive.

For Allen, the episode was traumatising. “I was just eating chips with my friends,” he said, reflecting on the sudden escalation. His account has attracted widespread attention online, serving as a cautionary example of how technological failures can have profound personal consequences and underscoring calls for stronger human oversight.