# Law-Enforcement AI-Assisted Analytics **Law-enforcement AI-assisted analytics** means computational tools used to surface patterns, links, narratives, relationships, and possible cross-jurisdiction connections for human investigators. ACIN's public materials say its AI helps "identify patterns, connect related incidents, and improve investigative efficiency across jurisdictions" ([[ACIN AI Fact Sheet]], p. 1). The final package adds executed service descriptions and policy materials, but not a completed deployment or performance audit. ## How it appears in the corpus - ACIN's fact sheet identifies related-incident detection, trend detection, and possible connections among "people, vehicles, locations, and events"; it says arrests, charging, and operational decisions remain human-controlled ([[ACIN AI Fact Sheet]], p. 1). - Peregrine's executed package describes comprehensive search, real-time alerts, dashboards, trend, geospatial, link, and temporal analysis. Its Arkansas addendum treats AI/ML as decision support subject to human review ([[Peregrine and Carahsoft Executed Contract]], `CON-002_Peregrine_Carahsoft_Executed_Contract_2026-03-17.pdf`, pp. 5, 22). - LeadsOnline's executed package provides AI-driven narrative sections, automated case summaries and timelines, cross-case links, phone-extraction processing, social-media analysis, communications, movement, GPS, gunfire, and LPR-hit analysis ([[LeadsOnline Executed Contract]], `CON-001_LeadsOnline_Executed_Contract_2026-07-13.pdf`, pp. 15-17, 46-47). - The statewide acceptable-use policy requires human review, approved tools, auditability, training, and security controls; the ACIN model policy calls for legal, privacy, procurement, vendor-risk, accuracy, bias, and retention review before acquisition or material change ([[ACIN AI Policy and CLEST Training Packet]], `GOV-202_Arkansas_AI_Acceptable_Use_Policy_Final_2026-02-26.pdf`, pp. 1-4; `GOV-002_Arkansas_Model_AI_Policy_TEMPLATE.pdf`, pp. 1-2). - The 55-page CLEST packet is prepared training material with blank external-approval fields; it is not evidence of CLEST approval or completed user training ([[ACIN AI Policy and CLEST Training Packet]], `TRN-001_ACIN_CLEST_AI_Training_Packet_FINAL_SOURCE_2026-07-06.pdf`, pp. 48-51). ## Stakeholders - [[Arkansas Criminal Intelligence Network]], its authorized users, and participating agencies. - [[Peregrine Technologies, Inc.]] and [[LeadsOnline]] as contracted analytics providers. - [[ACIN Oversight and Governance Board]], [[Arkansas Office of State Technology]], and [[Arkansas Commission on Law Enforcement Standards and Training]]. ## Notes The records establish intended functions, contractual capabilities, and written safeguards. They do not establish which model or feature was activated, what data it processed, its measured accuracy or bias, whether required reviews were completed, or how human-review rules operated in practice. The difference between the broad analytical capability and ACIN's non-surveillance classification remains open at [[T014 - ACIN Analytical Reach vs Non-Surveillance Classification]]; the conflict over whether investigative files are within scope is tracked at [[T030 - ACIN Incident-Report-Only Scope vs Investigative Data Services]].