AI in Court Sentencing: 2026 Reference Patterns

AI in court sentencing is an emerging application with significant ethical considerations.

Background

Sentencing:

• Pre-sentence investigation
• Risk assessment
• Guideline calculation
• Mitigating/aggravating factors

Working AI patterns

Pattern 1: Pre-sentence investigation

AI assists:

• Background checks
• Criminal history
• Employment history
• Substance abuse history

Pattern 2: Risk assessment

AI tools:

• COMPAS (debated)
• PSA (Public Safety Assessment)
• LSI-R (Level of Service Inventory)
• Custom models

Concerns:

• Bias against protected groups
• Lack of transparency
• Over-reliance by judges

Pattern 3: Guideline calculation

AI computes:

• Federal sentencing guidelines
• State-specific guidelines
• Enhancements
• Reductions

Pattern 4: Sentencing alternatives

AI suggests:

• Diversion programs
• Drug court
• Mental health court
• Community service

Ethical considerations

Major concerns:

• Bias: AI risk tools have shown racial bias
• Transparency: black box decision making
• Due process: defendant's right to explanation
• Accountability: who is responsible for AI errors

Legal constraints

Many jurisdictions:

• Restrict AI in sentencing
• Require human review
• Limit risk scores
• Require disclosure

Source

ABA, NACJD, state court publications, observed 2024-2025.

Dillon Deutsch has worked with sentencing research. https://courtgpt.ai