Finland Police's Facial Recognition Trial to Identify Sexual Abuse Victims Deemed Illegal

Finland's National Police Board was reprimanded for illegal processing of special categories of personal data in a facial recognition trial to identify potential victims of child sexual abuse.

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 involved the unlawful processing and transmission of sensitive biometric facial data to an unvetted third-party platform without the consent of the data subjects or authorization from the data controller.

Additional risk subdomains

• 6.5 Governance failure: The police unit tested and used the AI tool independently without the approval or oversight of the National Police Board, highlighting a failure in internal governance and regulatory compliance.

Causal factors

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

The privacy incident was caused by the human decision of NBI officers to use an unapproved third-party tool, resulting in an unintentional personal data breach while pursuing the goal of victim identification.

EU AI Act risk tier

• Risk tier: 2 High Risk

Risk Level 2 (High Risk). The system is a facial recognition application used in law enforcement, which is explicitly classified as high-risk under the EU AI Act due to its significant implications for fundamental rights and privacy.

AI system and alleged parties

• AI system: Clearview AI
• AI purpose: Face Recognition; Identification
• Behaviour type: Tool
• Alleged developer: Clearview AI
• Alleged deployer: Finland National Bureau of Investigation
• Alleged harmed parties: Finland National Bureau of Investigation

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 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 Minor, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Privacy

Reported: Yes, the report explicitly describes privacy violations and a personal data breach.

Directly caused: The NBI unit performed around 120 searches using Clearview AI, uploading biometric facial images of individuals to a third-party service without ensuring data protection compliance or knowing how the data would be stored or shared.

Inferred additional harm: It is likely that the uploaded biometric data remains stored on Clearview AI's platforms, potentially exposing the affected individuals to ongoing privacy risks.

People affected

• Occurrences reported: 1
• People reportedly harmed: 120
• People reportedly exposed: 120

Potential causes

Management

• Lack of Internal Oversight: NBI unit made independent decision to use tool without management approval.
• Failure to Monitor Staff: National Police Board failed to prevent unauthorized software trials.
• Poor Internal Communication: NBI leadership was unaware of tool usage until contacted by media.

Technology

• Misleading Software Capabilities: Belief that system does not store searches, leading to false sense of security.
• Unregulated Scraping Database: Clearview database relied on billions of scraped social media images.
• Potential Match Inaccuracies: Facial recognition software provides probabilities, not facts, risking errors.

Data Inputs

• Unsanctioned Biometric Processing: Processing of special categories of sensitive personal data without consent.
• Lack of Data Retention Details: Police did not determine how long data would be stored or if shared.
• Use of Real Victim Images: Real images of child exploitation victims were uploaded to a third-party app.

Human Factors

• Unit Overconfidence in Tool: Staff assumed the tool was safe based on a brief presentation at Europol.
• Workload Pressure: Unit sought automation to control increased child abuse screening workload.

Process and Methods

• Lack of Prior Impact Assessment: No data protection impact assessment was conducted before starting the trial.
• Unchecked Software Sourcing: Free trials were used without formal procurement or security vetting.
• Inadequate Training on Rules: Personnel were not sufficiently familiar with data processing regulations.

Regulatory Environment

• Ambiguous Law Enforcement Rules: Unclear boundaries regarding GDPR exemptions for covert police investigations.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide clear, consistent, and official details from both the Finnish police and the European Data Protection Board regarding the unauthorized trial, the number of searches, and the regulatory reprimand.

The unauthorized use of Clearview AI by the Finnish National Bureau of Investigation represents a minor national security concern characterized by internal governance failures and regulatory non-compliance. While the incident involved the unlawful processing of biometric data of approximately 120 individuals, it was a localized law enforcement trial with no broader threat to national sovereignty or critical infrastructure.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: Finland
• Alleged perpetrator: National Bureau of Investigation of Finland

Threat characteristics

• Imminence: Long-term. Represents an ongoing governance and policy challenge regarding unauthorized AI procurement rather than an active crisis.
• Autonomy: Human-controlled. The AI tool was manually operated by police officers to perform specific searches and did not execute decisions autonomously.
• Novelty: Established threat. Unauthorized law enforcement use of commercial facial recognition software is a well-documented issue globally.

Impact by dimension

• Physical security: Negligible. No physical systems, critical infrastructure, or human safety threats were involved in this regulatory compliance incident.
• Information security: Minor. Unsanctioned transmission of biometric data of approximately 120 individuals to a foreign commercial database represents a minor data security compromise.
• Sovereignty: Minor. Minor impact on government functions due to unauthorized tool adoption by a law enforcement unit bypassing internal oversight and national data protection regulations.
• Economic security: Negligible. No threats to financial systems, strategic industries, or national economic and technological competitiveness were identified.
• Societal stability: Minor. Limited privacy and human rights impact involving the unauthorized processing of biometric data for 120 individuals, resolved through standard regulatory reprimands.