Predictive Policing Program by Florida Sheriff’s Office Allegedly Violated Residents’ Rights and Targeted Children of Vulnerable Groups

The Intelligence-Led Policing model rolled out by the Pasco County Sheriff’s Office was allegedly developed based on flawed science and biased data that also contained sensitive information and irrelevant attributes about students, which critics said to be discriminatory.

Risk classification

• Primary risk domain: 1 Discrimination & Toxicity
• Primary risk subdomain: 1.1 Unfair discrimination and misrepresentation

The ILP algorithm systematically targeted individuals, including minors and people with developmental disabilities, based on biased scoring rubrics, leading to unequal treatment and unfair outcomes.

Additional risk subdomains

• 4.1 Disinformation, surveillance, and influence at scale: The system was used to conduct systematic, automated, and coordinated surveillance of over 1,000 residents and 20,000 students.
• 2.1 Compromise of privacy by obtaining, leaking or correctly inferring sensitive information: The program obtained and utilized confidential student records, including grades, attendance, and child welfare histories, without parental consent.

Causal factors

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

The incident was caused by the deployment of the ILP algorithm, which was intentionally designed and used by the Sheriff's Office to target and harass specific residents.

EU AI Act risk tier

• Risk tier: 1 Unacceptable

Unacceptable Risk: The system represents a form of social and criminal scoring by a government authority that targets vulnerable groups based on age and disability, which is prohibited under fundamental rights.

AI system and alleged parties

• AI system: None named
• AI purpose: Predictive Policing; Recidivism Prediction
• Behaviour type: Assistant
• Alleged developer: unknown
• Alleged deployer: Pasco Sheriff's Office
• Alleged harmed parties: Pasco students with disabilities, Pasco residents, Pasco Black students

Harm severity

Highest direct severity in any category: Substantial. 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 Substantial
• 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 Minor
• Differential treatment: direct Substantial, indirect Substantial
• Civil rights: direct Substantial, indirect Substantial
• Democracy: direct Substantial, indirect Substantial
• Privacy: direct Negligible, indirect Substantial
• Psychological: direct Negligible, indirect Substantial
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Physical

Reported: The report explicitly describes physical harm, including a suicide and a physical collapse requiring hospitalization.

Directly caused: N/A

Indirectly caused: A 14-year-old target, Matthew, committed suicide by prescription drug overdose after continuous police checks. Another target, Rio Wojtecki, collapsed on the bathroom floor due to anxiety and was hospitalized.

Inferred additional harm: It is likely that other targeted individuals experienced physical altercations or health crises during the 12,500+ warrantless police visits, but specific details are not provided.

Financial

Reported: The report explicitly describes financial losses from heavy fines and loss of employment.

Directly caused: N/A

Indirectly caused: Targets were saddled with heavy fines, such as a $2,500 fine for backyard chickens and $100 for code violations. Heilman became a convicted felon and was unable to find work for two years. Billings' arrest record dashed his dream of law school.

Inferred additional harm: Many of the 1,000 targeted individuals likely suffered significant financial losses from legal fees, court costs, bail, and lost employment opportunities due to retaliatory arrests and citations.

Malicious content

Reported: The report explicitly describes the creation and spread of malicious content by the Sheriff's Office to humiliate targets.

Directly caused: N/A

Indirectly caused: The Sheriff's Office posted details of Robert Billings III's arrest on Facebook, and posted a crying man's photo as 'SAD CRIMINAL OF THE DAY' to ridicule him.

Inferred additional harm: N/A

Differential treatment

Reported: The report explicitly describes systematic differential treatment of individuals and groups based on algorithmic scores.

Directly caused: The algorithm scored and flagged individuals as 'prolific offenders' or 'at-risk,' subjecting them to intense police scrutiny and harassment that other residents did not face.

Indirectly caused: Juveniles and individuals with developmental disabilities (like autism) were disproportionately targeted for aggressive policing and pushed into the school-to-prison pipeline.

Inferred additional harm: Socio-economically deprived families and marginalized individuals were likely subjected to systemic bias and unequal treatment under the guise of objective data-driven policing.

Civil rights

Reported: The report explicitly describes violations of First, Fourth, and Fourteenth Amendment rights.

Directly caused: The algorithm's targeting lists led directly to warrantless searches, continuous harassment, and retaliatory arrests, violating residents' constitutional rights to be secure in their homes and receive due process.

Indirectly caused: The school district shared confidential student data (grades, discipline, abuse history) with police without parental consent, violating student privacy rights.

Inferred additional harm: Systemic civil rights violations likely affected many of the 1,000+ individuals ensnared in the program, who had no legal mechanism to challenge their status on the list.

Democracy

Reported: The report explicitly describes the erosion of democratic norms, including the presumption of innocence.

Directly caused: The predictive policing system eroded the fundamental democratic principle of 'innocent until proven guilty' by treating individuals as 'predicted guilty' based on algorithmic scores.

Indirectly caused: The Sheriff's Office used its political connections and power to avoid oversight, retaliate against internal critics, and maintain absolute control, weakening local democratic accountability.

Inferred additional harm: The lack of regulatory frameworks and oversight for predictive policing algorithms poses a broader threat to democratic governance and the rule of law in local communities.

Privacy

Reported: The report explicitly describes privacy violations involving the unauthorized sharing of sensitive student records.

Directly caused: N/A

Indirectly caused: The school district shared sensitive, personally identifiable information (grades, attendance, discipline, and abuse histories of 20,000 students) with the Sheriff's Office without parental knowledge or consent.

Inferred additional harm: The algorithm's reliance on a vast database of police reports, Facebook pages, bank statements, and property records likely violated the privacy expectations of thousands of residents.

Psychological

Reported: The report explicitly describes severe anxiety, fear, and psychological pressure experienced by targeted individuals and their families.

Directly caused: N/A

Indirectly caused: Rio Wojtecki (15) suffered severe anxiety and collapsed. Matthew (14) experienced extreme psychological pressure and suicidal thoughts before taking his life. Children of targeted families hid under beds in fear as police banged on windows.

Inferred additional harm: Widespread psychological trauma, anxiety, and distress are highly likely among the 1,000+ individuals and their families who were subjected to continuous, aggressive police harassment.

People affected

• Occurrences reported: 1
• People reportedly harmed: 1000
• People reportedly exposed: 20000

Potential causes

Management

• Prioritizing Metrics over Rights: Focusing on property crime reduction stats at the expense of rights.
• Retaliation Against Critics: Internal management pushed out deputies who questioned the program.
• Defiant Leadership: Leadership refused to acknowledge flaws or apologize for harassment.

Technology

• Flawed Scoring Algorithm: Algorithm assigns points for arrests and suspect status, creating bias.
• Lack of Algorithmic Transparency: The exact weighting of scores and enhancement criteria remains hidden.
• Pre-criminal Predictive Modeling: Unscientifically attempts to predict future crimes before they occur.

Data Inputs

• Biased Arrest Records: Uses arrest histories regardless of whether charges were dropped.
• School Data Sharing: Mines school grades, attendance, and child-welfare histories.
• Unverified Suspect Information: Includes subjective police reports and witness designations as inputs.

Human Factors

• Aggressive Deputy Enforcement: Deputies ordered to harass targets and make their lives miserable.
• Confirmation Bias of Analysts: Analysts manually select targets, reinforcing algorithmic bias.
• Overreliance on Scores: Officers trust generated scores to justify warrantless home visits.

Process and Methods

• Lack of Appeal Mechanism: No way for residents to check their status or appeal being listed.
• Misapplied Code Enforcement: Using minor code violations as a weapon to force cooperation.
• Siloed Information Sharing: Sharing student data between schools and police without consent.

Regulatory Environment

• No Predictive Policing Laws: Absence of state or federal laws banning or regulating pre-crime tools.
• Weak Privacy Enforcement: Student privacy laws failed to prevent sharing of sensitive records.
• Inadequate Independent Oversight: Sheriffs operate with vast autonomy and little external accountability.

Information quality

• Classification confidence: High
• Reason for confidence: The investigative reporting by the Tampa Bay Times is highly detailed, referencing official manuals, body-camera footage, court filings, and interviews with both victims and former deputies. The Sheriff's Office also provided extensive public rebuttals, ensuring both perspectives are documented.
• Ambiguities identified: The exact mathematical weights of the scoring algorithm are not fully disclosed in the public manuals.
• Alternative interpretations: The Sheriff's Office argues the program is standard 'Intelligence-Led Policing' focused on recidivism reduction rather than 'Predictive Policing' or 'pre-crime' targeting.

A local law enforcement agency's deployment of a predictive policing algorithm systematically violated the civil rights of over 1,000 residents and exposed 20,000 students to unauthorized surveillance. The program's severe societal impacts and constitutional violations prompted federal oversight, illustrating the national security risks of unregulated domestic algorithmic surveillance.

• Overall national security impact: Substantial
• Response level: Substantial
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Pasco County Sheriff's Office

Threat characteristics

• Imminence: Long-term. The program was a long-term strategic deployment over several years rather than an immediate crisis requiring urgent national response.
• Autonomy: Human-supervised. The algorithm generated the targeting lists and scores, but human deputies made the final decisions to execute searches, visits, and arrests.
• Novelty: Evolved capability. Represents an evolved and highly aggressive integration of non-traditional data sources like school and child welfare records into predictive policing systems.

Impact by dimension

• Physical security: Negligible. No threats to physical critical infrastructure, kinetic weapons, or military systems identified in the incident.
• Information security: Minor. While involving unauthorized access to student records, it was a local initiative rather than a national-level intelligence compromise or foreign information operation.
• Sovereignty: Minor. Local government and law enforcement processes were compromised by algorithmic bias, but this did not threaten national sovereignty or federal operations.
• Economic security: Negligible. No significant threats to national economic stability, critical supply chains, or strategic technological competitive advantage.
• Societal stability: Substantial. Systematic violation of constitutional rights targeting over 1,000 residents and 20,000 students, prompting federal civil rights investigations and causing severe societal harm.