Maryland Police Allegedly Relied on Facial Recognition Lead in Wrongful Arrest and Detention of Kimberlee Williams

Kimberlee Williams was reportedly arrested in Oklahoma on Maryland warrants after police allegedly relied on a facial recognition lead that incorrectly identified her as a suspect in bank fraud cases. The ACLU said Williams had never been to Maryland, police failed to investigate alibi evidence placing her in Oklahoma, and she spent about six months in jail before the charges were dropped.

Risk classification

• Primary risk domain: 1 Discrimination & Toxicity
• Primary risk subdomain: 1.3 Unequal performance across groups

The facial recognition technology exhibits unequal performance across demographic groups, producing higher false match rates for people of color, women, and the elderly, which directly led to disproportionate wrongful arrests.

Additional risk subdomains

• 7.3 Lack of capability or robustness: The AI system failed to perform reliably, generating a false match that misidentified an innocent person as a suspect.
• 5.1 Overreliance and unsafe use: Police officers overrelied on the facial recognition lead, failing to conduct independent investigations or verify the suspect's identity.

Causal factors

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

The incident was initiated by an incorrect match generated by a deployed facial recognition AI system, which was an unintentional failure of the technology.

EU AI Act risk tier

• Risk tier: 2 High Risk

Risk Level 2. High Risk: AI systems that have significant implications for safety, fundamental rights, and public interests. Examples include law enforcement applications, including facial recognition.

AI system and alleged parties

• AI system: facial recognition technology
• AI purpose: Face Recognition; Identification
• Behaviour type: Assistant
• Alleged developer: Unknown facial recognition technology developers
• Alleged deployer: Prince George's County Police Department, Montgomery County Police Department, Anne Arundel County Police Department
• Alleged harmed parties: Wrongfully arrested individuals, Kimberlee Williams, Epistemic integrity

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Minor. 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 Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Minor
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Minor, 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

Financial

Reported: Yes, the report mentions that [PERSON_002] lost her job due to the arrest and was released with no money.

Indirectly caused: [PERSON_002] suffered significant financial loss, including losing her job due to being jailed for six months, and was left with no money upon her release.

Inferred additional harm: The other 13 wrongfully arrested individuals likely suffered substantial financial losses from lost wages, legal fees, and potential job loss during their detentions.

Differential treatment

Reported: Yes, the report explicitly describes how facial recognition technology produces higher false match rates for people of color, women, younger people, and the elderly.

Directly caused: The AI system directly exhibited differential treatment by performing less accurately for specific demographic groups, leading to a higher rate of false matches for these populations.

Indirectly caused: This technical bias resulted in a disproportionate number of Black individuals being wrongfully arrested and subjected to police scrutiny.

Inferred additional harm: It is likely that systemic bias in facial recognition deployments continues to cause unequal police targeting and higher rates of false identifications for minority groups nationwide.

Civil rights

Reported: Yes, the report explicitly describes wrongful arrests and false imprisonments resulting from reliance on the technology.

Indirectly caused: At least 14 individuals had their civil rights violated through wrongful arrests and subsequent detentions, including [PERSON_002] who was jailed for six months without cause.

Inferred additional harm: Many other individuals may have had their civil liberties compromised or faced unlawful detentions due to undisclosed police reliance on erroneous facial recognition matches.

Psychological

Reported: Yes, the report describes significant distress, anxiety, and trauma resulting from wrongful arrest and detention.

Indirectly caused: The wrongful arrest and subsequent six-month jail detention caused severe emotional distress, anxiety, and ongoing fear of law enforcement for [PERSON_002] and her family.

Inferred additional harm: It is highly likely that the other 13 individuals wrongfully arrested also suffered severe psychological trauma, anxiety, and distress from being falsely imprisoned.

People affected

• Occurrences reported: 14
• People reportedly harmed: 14
• People reportedly exposed: 14

Potential causes

Management

• Failure to Enforce Policies: Management failed to enforce policies stating FRT is not grounds for arrest.

Technology

• High False Match Rates: FRT often produces false matches, especially for women and minorities.
• Inherent Demographic Bias: Studies show FRT has higher error rates for people of color and women.

Data Inputs

• Low-Quality Security Images: Using low-quality security camera footage led to inaccurate FRT matches.
• Lack of Database Context: Matches sent via listserv lacked context on how search was conducted.

Human Factors

• Confirmation Bias in Officers: Officers visually confirmed false matches due to cognitive bias from FRT.
• Overreliance on Automation: Police treated unverified FRT leads as sufficient proof for arrest.

Process and Methods

• Inadequate Alibi Verification: Detectives failed to verify alibis or investigate other obvious leads.
• Flawed Photo Lineup Practices: Using FRT leads directly in lineups biased witnesses toward false matches.
• Concealment of FRT Use: Police hid their reliance on FRT from courts during warrant applications.

Regulatory Environment

• Absence of Use Restrictions: Lack of legal bans or restrictions on police use of FRT in Maryland.

Information quality

• Classification confidence: High
• Reason for confidence: The report provides clear, detailed, and consistent accounts of the wrongful arrest of [PERSON_002] and lists 13 other specific cases. The role of the facial recognition technology is explicitly stated, and the resulting harms (wrongful arrest, job loss, emotional distress) are directly documented.
• Ambiguities identified: The specific developer or brand of the facial recognition software used is not named in the report.
• Alternative interpretations: None. The causal link between the false AI match and the subsequent wrongful arrest is clear and undisputed.

An innocent individual was wrongfully jailed for six months due to an incorrect facial recognition match by Maryland police. This incident highlights ongoing domestic civil rights and societal stability concerns regarding the overreliance on biased facial recognition technology in law enforcement, though it presents negligible direct threat to national security.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Maryland police

Threat characteristics

• Imminence: Long-term. Represents an ongoing systemic issue with law enforcement technology deployment rather than an active, immediate crisis.
• Autonomy: Human-controlled. The AI system acts as an investigative assistant, requiring human officers to verify leads and make the final arrest decisions.
• Novelty: Established threat. Multiple similar incidents of wrongful arrests due to facial recognition errors have been documented previously in the United States.

Impact by dimension

• Physical security: Negligible. No threat to physical infrastructure, military systems, or physical safety of national scale.
• Information security: Negligible. No evidence of intelligence capabilities compromise, espionage, or systematic information warfare.
• Sovereignty: Minor. Minor impact on local judicial and law enforcement processes due to faulty automated leads, but manageable within standard legal procedures.
• Economic security: Negligible. No significant threat to national economic stability or strategic technological competitive advantage.
• Societal stability: Minor. Flawed facial recognition led to the wrongful arrest and six-month detention of an innocent individual, highlighting systemic civil rights concerns and demographic biases.