Law Enforcement AI Accountability Act of 2025 (LEA³)
A BILL
To establish national standards, oversight, and restrictions on the use of artificial intelligence in law enforcement to protect civil liberties, eliminate discriminatory practices, and restore public trust.
SECTION 1. SHORT TITLE
This Act may be cited as the “Law Enforcement AI Accountability Act of 2025” or “LEA³.”
SECTION 2. FINDINGS
Congress finds the following:
AI and algorithmic systems are increasingly used in law enforcement for surveillance, predictive policing, facial recognition, and risk assessment.
Without clear federal standards, these systems have produced biased outcomes, constitutional violations, and erosions of public trust.
Law enforcement must not use AI to circumvent rights protected under the U.S. Constitution, including the Fourth, Fifth, and Fourteenth Amendments.
Civilian oversight, transparency, and public accountability are essential for any AI system used in policing.
SECTION 3. DEFINITIONS
As used in this Act:
Law Enforcement AI refers to any AI system used by local, state, or federal law enforcement agencies, including surveillance, risk analysis, predictive tools, and biometric identification.
Civilian Oversight Board (COB) means an independent public body with legal authority to review, approve, and audit AI usage in policing.
Prohibited AI Use means any application that violates constitutional rights, targets protected classes, or lacks transparency and due process.
SECTION 4. NATIONAL STANDARDS FOR LAW ENFORCEMENT AI
(a) All AI systems used in U.S. law enforcement must:
Be certified by the Federal AI Constitutional Authority (FACA)
Comply with the Five Codex Laws
Be fully auditable and subject to civilian review
Include documented logic, datasets, and decision outputs
(b) Certification must be renewed every 18 months and is subject to revocation for misuse, bias, or breach.
SECTION 5. PROHIBITED USES
(a) Law enforcement agencies are prohibited from using AI in the following ways:
Predictive policing systems targeting individuals based on race, geography, or social network inference
Real-time facial recognition in public without probable cause and judicial warrant
AI-generated risk scores in sentencing or bail without human judicial override
Surveillance systems that log, profile, or categorize constitutionally protected activities (e.g., protests, religious gatherings, political events)
(b) Any agency found using prohibited AI systems shall be subject to civil penalties and loss of federal funding.
SECTION 6. MANDATORY CIVILIAN OVERSIGHT
(a) All jurisdictions using Law Enforcement AI must establish or partner with an approved Civilian Oversight Board (COB).
(b) COBs must:
Include diverse, independent community representatives
Have the authority to review procurement, audits, and incident data
Publish public annual reports
Hold hearings on AI usage complaints or incidents
(c) COB findings must be reported to FACA for national accountability tracking.
SECTION 7. DATA PRIVACY AND REDRESS
(a) Individuals shall have the right to:
Be notified when AI is used against them in law enforcement contexts
Request access to records about that use
Appeal or challenge AI-generated decisions through a formal redress process
(b) Data used in AI training or decision-making must be anonymized, bias-tested, and free from illicit surveillance sources.
SECTION 8. FUNDING
(a) This Act shall be funded through allocations from the American Reboot Act.
(b) No new taxes shall be levied.
(c) Federal funding to law enforcement agencies is contingent on compliance.
SECTION 9. RULEMAKING AND ENFORCEMENT
(a) FACA shall issue regulatory guidance within 12 months.
(b) DOJ and FACA may jointly investigate violations and recommend corrective actions, penalties, or federal funding suspensions.
SECTION 10. SEVERABILITY
If any provision of this Act is held invalid, the remainder shall remain in effect.
