Flock Safety License-Plate Reader Reportedly Misread Plate, Leading Toledo, Ohio, Police to Stop Motorist at Gunpoint and Deploy K-9

Toledo, Ohio, police stopped motorist Brandon Upchurch at gunpoint after a Flock Safety license-plate reader incorrectly matched his truck to a stolen plate. Officers released a K-9 that bit him and arrested him. Police later said the alert was a misread; one officer was reprimanded for failing to verify the plate, and all charges against Upchurch were later dropped.

Risk classification

• Primary risk domain: 7 AI system safety, failures, & limitations
• Primary risk subdomain: 7.3 Lack of capability or robustness

The primary failure was the AI system's lack of robustness in accurately reading license plate characters under real-world conditions, misidentifying a '7' as a '2' and triggering a false police response.

Additional risk subdomains

• 5.1 Overreliance and unsafe use: Police officers overrelied on the automated alert and initiated a high-risk, armed intervention without performing the mandatory visual verification of the license plate.

Causal factors

• Entity: AI
• Intent: Unintentional
• Timing: Post-deployment

The physical and legal harms resulted from an unexpected character misidentification error by the Flock Safety AI system after it was deployed in the field.

EU AI Act risk tier

• Risk tier: 2 High Risk

High Risk: The AI system is used in law enforcement applications to identify vehicles of interest, which has significant implications for public safety, personal liberty, and fundamental human rights.

AI system and alleged parties

• AI system: Flock Safety license-plate reader (Flock Safety)
• AI purpose: License Plate Recognition; Threat Detection
• Behaviour type: Tool
• Alleged developer: Surveillance technology developers, Flock Safety, Automated license plate reader (ALPR) developers
• Alleged deployer: Toledo Police Department, Police departments in the United States, Law enforcement, Adrian Wilson
• Alleged harmed parties: Road users, Motorists, Individuals mistakenly detained by law enforcement, Epistemic integrity, Drivers, Brandon Upchurch

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Negligible. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Minor
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Minor
• Financial: direct Negligible, indirect Minor
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Negligible, indirect Minor
• Civil rights: direct Negligible, indirect Minor
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Minor
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Physical

Reported: The report explicitly describes severe physical injuries to [PERSON_001] caused by a police dog deployment during the stop.

Indirectly caused: The victim suffered seven severe dog bites to his forearm and elbow, lost feeling in his fingers, and experienced permanent physical limitations that prevented him from continuing his forklift and lawn care work.

Inferred additional harm: The reports mention at least twelve other instances nationwide of people being stopped at gunpoint, jailed, or mauled by police dogs due to Flock misreads, suggesting additional unquantified physical injuries occurred in those other cases.

Property

Reported: The report explicitly describes property loss resulting from the financial strain of the incident.

Indirectly caused: The victim was forced to sell his truck to keep up with rent after losing his livelihood due to his physical injuries.

Financial

Reported: The report describes legal settlements and loss of livelihood resulting from the incident.

Indirectly caused: The victim lost his lawn care business and forklift job due to physical disability. The city of Toledo paid a $35,000 legal settlement to Upchurch, and other cities paid settlements such as $45,000 in Atherton, California.

Inferred additional harm: The victim was eventually evicted from his home due to an inability to pay rent, representing severe unquantified long-term financial ruin.

Differential treatment

Reported: The reports describe concerns raised by the NAACP regarding the historical context of police dog deployments against Black individuals.

Indirectly caused: The Toledo branch of the NAACP highlighted that the deployment of a K-9 on a Black man was reminiscent of civil rights era police violence, raising concerns of systemic differential treatment.

Inferred additional harm: If the system's error rates or police deployment patterns disproportionately affect minority neighborhoods where surveillance cameras are heavily concentrated, it could lead to systemic differential treatment.

Civil rights

Reported: The reports explicitly describe violations of civil rights, wrongful arrests, and excessive use of force.

Indirectly caused: The victim's civil rights were violated through an unjustified armed stop, physical mauling, and wrongful arrest/detention for hours on dismissed charges.

Inferred additional harm: The twelve other documented false positive stops nationwide represent systemic violations of Fourth Amendment rights against unreasonable search and seizure.

Psychological

Reported: The report explicitly describes severe psychological trauma and distress experienced by the victim and his family.

Indirectly caused: [PERSON_001] reported being deeply traumatized, developed a fear of dogs that forced him to give up his own pets, and a 3-year-old child in another cited incident was left terrified in a vehicle during a similar high-risk stop.

Inferred additional harm: The other eleven families and individuals subjected to high-risk felony stops at gunpoint due to system errors likely suffered severe acute stress and long-term psychological trauma.

People affected

• Occurrences reported: 12
• People reportedly harmed: 15
• People reportedly exposed: 15

Potential causes

Management

• Prioritizing Growth Over Accuracy: Vendor prioritized rapid expansion without disclosing operational error rates.
• Allowing Lower Confidence Searches: Management allowed officers to lower confidence thresholds, increasing errors.
• Inadequate Officer Training: Inadequate training on the limitations and verification protocols of ALPR tech.

Technology

• Character Misidentification: The Flock camera misread the character '7' on the license plate for a '2'.
• Database Sync Failure: Stolen status of vehicles already recovered was not updated in the system.
• Unreliable LPR Performance: System accuracy is impacted by dirty plates, covers, and lighting conditions.

Data Inputs

• Incorrect Camera Hit: The camera fed a false-positive stolen plate alert directly to the officers.
• Obscured License Plate: Dirt or covers on license plates caused the AI model to misread characters.

Human Factors

• Failure to Verify Alert: The officer did not visually verify the license plate before deploying force.
• Adrenaline and Information Overload: Inundation of automated pings increased officer adrenaline and rushed actions.
• Confirmation Bias: Officers assumed the suspect was fleeing instead of tripping during the stop.

Process and Methods

• High-Risk Felony Stop Protocols: Officers immediately drew guns and used high-risk tactics based on an ALPR hit.
• Inadequate K-9 Deployment Policy: Policy allowed canine deployment without confirming the severity of the crime.
• Lack of Real-Time Tag Running: Officers failed to run the plate through NCIC manually to confirm the hit.

Regulatory Environment

• Lack of Independent Efficacy Data: No independent regulatory body measures or standardizes LPR accuracy rates.
• Absence of Error-Rate Disclosure: No regulatory mandate forces vendors to publicize local system misread rates.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide highly detailed, consistent, and verified accounts of the specific incident in Toledo, including body camera transcripts, police department statements, and legal settlement details. The role of the Flock Safety camera in misidentifying the plate is explicitly documented and acknowledged by both the police department and the court.
• Ambiguities identified: None of significance regarding the primary incident; there is some minor ambiguity regarding the exact nationwide error rates of the Flock system as the company declines to publish specific data.
• Alternative interpretations: The incident could be viewed purely as a failure of police procedure (failure to visually verify), but the underlying trigger was an algorithmic misidentification, making it a clear AI-human interaction failure.

An automated license plate reader system misidentified a plate character, leading local police to detain an innocent driver at gunpoint and deploy a police dog. This incident is part of a nationwide pattern of false positive arrests and civil rights violations stemming from AI-driven mass surveillance, highlighting systemic risks to societal stability, civil liberties, and public trust in law enforcement.

• Overall national security impact: Substantial
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. The threat represents an ongoing systemic capability and policy concern regarding domestic surveillance rather than an active, imminent national security crisis.
• Autonomy: Human-supervised. The AI system operates autonomously to detect and match plates, but human officers are supposed to verify the alerts and make the final decision to initiate enforcement actions.
• Novelty: Evolved capability. The incident represents a significant advancement in the scale and integration of automated surveillance tools in domestic policing, leading to recurring systemic failures.

Impact by dimension

• Physical security: Minor. The incident involved physical harm and injuries to a civilian due to police dog deployment, but did not threaten critical infrastructure, national military assets, or cause mass casualties.
• Information security: Negligible. No foreign intelligence compromise, state-sponsored disinformation, or systematic information warfare operations were indicated in the incident.
• Sovereignty: Minor. Local law enforcement operations and municipal police functions were disrupted due to erroneous alerts, but there was no threat to state authority or federal constitutional processes.
• Economic security: Negligible. Economic impacts were limited to individual financial loss and local government legal settlements, with no broader strategic threat to national economic stability or key technological industries.
• Societal stability: Substantial. The widespread deployment of unverified ALPR systems has led to systemic civil rights violations, wrongful arrests of innocent citizens, and public pushback, threatening trust in public safety institutions.