This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
Police departments across the U.S. have utilized Clearview AI's facial recognition software to identify criminal suspects, frequently failing to disclose this usage to defendants. This practice has led to multiple wrongful arrests, particularly affecting people of color, and has raised significant concerns regarding due process, transparency, and the reliability of the technology in law enforcement.
Police departments across the U.S. have used facial recognition software to identify suspects in criminal investigations, leading to multiple false arrests and wrongful detentions. The software's unreliability, especially in identifying people of color, has resulted in misidentifications that were not disclosed to defendants. In some cases, individuals were unaware that facial recognition played a role in their arrest, violating their legal rights and leading to unjust detentions.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Accuracy and effectiveness of AI decisions and actions are dependent on group membership, where decisions in AI system design and biased training data lead to unequal outcomes, reduced benefits, increased effort, and alienation of users.
Also classified as
Law enforcement officers overrelied on the AI's facial recognition matches as the sole basis for arrests without conducting independent verification.
Police departments systematically concealed the use of facial recognition software from defendants, preventing them from challenging the technology's reliability.
The facial recognition technology exhibits unequal performance across demographic groups, with significantly higher false match rates for people of color, directly leading to the wrongful arrest of Black individuals like Quran Williams.
AI system
Due to a decision or action made by an AI system
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The incident was initiated by a false match generated by the Clearview AI system post-deployment, which was an unintentional and unexpected outcome of the system's use.
Added categories peak at 3: Substantial (Epistemic Harm, Indirect)
Widespread and undisclosed use of Clearview AI facial recognition software by U.S. police departments has led to wrongful arrests and systemic violations of civil rights and due process, particularly affecting people of color. While not a direct kinetic threat, the practice undermines judicial integrity, public trust, and constitutional protections on a substantial scale.
The AI system acts as an Assistant because it is designed to generate investigative leads and suggest potential matches from a database, requiring human verification and decision-making before any action is taken. The report notes that 'facial recognition was intended to serve merely as a lead-generating tool, not a basis for arrest.'
High Risk: The report describes the use of facial recognition technology in law enforcement, which is explicitly classified as a High Risk application under the EU AI Act due to its significant implications for safety and fundamental rights.
The primary goal of the Clearview AI system is to perform facial recognition by comparing probe images against a database of billions of scraped photos to identify individuals.
The reports provide detailed, corroborating accounts from reputable sources (biometricupdate.com and The Washington Post) regarding the specific case of Quran Williams and the broader systemic issue of undisclosed facial recognition use by police. The role of Clearview AI and the resulting harms (wrongful arrests, lack of due process, racial bias) are explicitly documented with quotes from legal filings, police records, and official statements.
None. The facts of the misidentification and subsequent civil settlement are clear and uncontested.
Technology
Algorithmic Racial Bias
FRT has higher false match rates for people of color, causing misidentification.
Lack of Algorithmic Reliability
The software does not consistently produce accurate results on probe images.
Data Inputs
Low-Quality Surveillance Footage
Poor crime scene images fed into database lead to unreliable search matches.
Scraped Social Media Databases
Massive databases contain irrelevant images, increasing false positive risks.
Human Factors
Cognitive Bias in Matching
Investigators blindly accepted the AI match without looking for discrepancies.
Lack of Operator Training
Officers lacked training on how to interpret and verify software results.
Process and Methods
Omission of AI Use in Affidavits
Officers hid FRT use by citing 'a credible source' in arrest warrants.
Lack of Corroborating Investigation
No basic verification, such as checking location or physical features, occurred.
Siloed Information Sharing
Baton Rouge detective borrowed the unverified match to secure another warrant.
Management
Lack of Formal Agency Policies
JPSO failed to establish formal policies to guide facial recognition use.
Failure of Internal Oversight
Management failed to audit the misuse of FRT as a sole basis for arrests.
Regulatory Environment
Absence of Federal FRT Laws
No federal laws regulate the use of facial recognition by law enforcement.
Fragmented Local Ordinances
Inconsistent local regulations allow police to bypass local bans covertly.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
