Miami Police Deployed Facial Recognition to Arrest George Floyd Protestor Allegedly without Cause

Miami Police's arrest report for a George Floyd protestor did not disclose use of facial recognition, which allegedly did not meet the legal threshold for probable cause for arrest.

Risk classification

• Primary risk domain: 2 Privacy & Security
• Primary risk subdomain: 2.1 Compromise of privacy by obtaining, leaking or correctly inferring sensitive information

The incident involves the police using Clearview AI to scrape public social media data and match it to surveillance footage, thereby identifying a private individual without her consent, violating privacy expectations.

Additional risk subdomains

• 1.3 Unequal performance across groups: The reports highlight concerns that facial recognition technology has unequal accuracy and is programmatically less effective for identifying people of color, women, and children.
• 5.1 Overreliance and unsafe use: Police relied on the AI system's identification to make an arrest while omitting its use from the official arrest report, bypassing standard legal oversight.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The incident was caused by the intentional decision of law enforcement officers to deploy and use the Clearview AI facial recognition system post-deployment without disclosing its use in the arrest report.

EU AI Act risk tier

• Risk tier: 2 High Risk

High Risk: The report describes a facial recognition system used by law enforcement to identify a suspect. According to the EU AI Act definitions, 'Law enforcement applications, including facial recognition' are classified as High Risk (Level 2).

AI system and alleged parties

• AI system: Clearview AI
• AI purpose: Face Recognition; Identification
• Behaviour type: Assistant
• Alleged developer: Clearview AI
• Alleged deployer: Miami Police Department
• Alleged harmed parties: Oriana Albornoz, George Floyd protest participants

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

People affected

• Occurrences reported: 1
• People reportedly harmed: 1
• People reportedly exposed: 1

Potential causes

Management

• Lack of Organizational Oversight: Police departments deploy AI without transparent reporting or public consent.
• Inadequate Risk Assessment: Failure to assess the ethical and legal risks of deploying scraping tools.
• Prioritization of Speed over Safety: Prioritizing fast arrests over the legal rights of peaceful protesters.

Technology

• Algorithmic Demographic Bias: Algorithms are less effective for identifying women and people of color.
• Facial Recognition False Matches: Civilian-built AI tool misidentified an ex-policeman as a violent officer.
• Off-the-shelf Model Availability: Easy access to computer vision models allows quick creation of tracking tools.

Data Inputs

• Unconsented Social Media Scraping: Clearview AI and civilian bots scraped public photos to build face libraries.
• Low-quality Surveillance Footage: Low-resolution protest footage requires AI enhancement, increasing error risk.
• Lack of Verified Training Data: AI models trained on biased datasets produce inaccurate real-world matches.

Human Factors

• Omission of AI in Arrest Reports: Police omitted Clearview AI use in reports, hiding the technology's role.
• Potential for Personal Abuse: Officers routinely abuse confidential databases to spy on exes and journalists.
• Over-reliance on AI Matches: Using unverified AI matches without proper human validation of results.

Process and Methods

• No Probable Cause Verification: Using facial recognition results as sole identification without legal warrants.
• Lack of Audits and Minimization: No standardized procedures for data minimization or usage audits across states.
• Absence of Verification Protocols: Lack of rigorous human-in-the-loop validation before releasing matches.

Regulatory Environment

• Absence of Statewide Regulations: No statewide rules govern police use of facial recognition in Florida.
• Weak Local Bans and Loopholes: City-level bans do not stop federal agencies from using facial recognition.
• Inadequate Legislative Guardrails: Legislators are hesitant to put limits on policing tools post-9/11.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide consistent, detailed accounts of the specific incident involving Oriana Albornoz, the Miami Police Department, and Clearview AI. The facts regarding how the AI was used, the lack of disclosure in the arrest report, and the subsequent discovery by journalists are well-documented across multiple sources.
• Ambiguities identified: There is some ambiguity regarding whether the police actively used the software to monitor peaceful protesters or only applied it after a crime was alleged to have occurred, as the policy boundaries are described as a gray area.
• Alternative interpretations: The incident could be interpreted primarily as a police misconduct or transparency issue rather than an inherent failure of the AI system itself.

The Miami Police Department used Clearview AI facial recognition to identify a protester without disclosing its use in the arrest report. While raising significant civil liberties, privacy, and domestic policy concerns regarding biometric surveillance, the incident has negligible direct national security implications.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Miami Police Department

Threat characteristics

• Imminence: Long-term. Represents an ongoing debate and strategic concern over privacy and surveillance rather than an imminent national security crisis.
• Autonomy: Human-controlled. The AI system served as an investigative aid to match images, with human officers validating the results and making the final arrest decision.
• Novelty: Evolved capability. Represents an evolved capability in law enforcement surveillance through mass-scale scraping of public social media photos.

Impact by dimension

• Physical security: Negligible. No physical systems, critical infrastructure, or kinetic targeting capabilities were compromised or deployed in this local law enforcement incident.
• Information security: Negligible. No classified information, intelligence collection methods, or foreign disinformation campaigns were involved in this domestic law enforcement case.
• Sovereignty: Negligible. The incident involves local law enforcement procedures and does not threaten state authority, territorial control, or federal government functions.
• Economic security: Negligible. No threats to strategic industries, financial systems, or national economic competitive advantage were identified.
• Societal stability: Minor. Highlights domestic civil rights and privacy concerns regarding undisclosed law enforcement use of biometric surveillance, but represents a minor localized impact.