AI to Be Regulated in Indonesia's Criminal Procedure: What Government Teams Need to Prepare
Indonesia is moving to formally regulate the use of artificial intelligence in criminal procedure. Law Minister Supratman Andi Agtas confirmed a Draft Presidential Regulation-deriving from Law Number 20 of 2025 on the Criminal Procedure Code (KUHAP)-will set the rules for AI within an information technology-based criminal justice system.
Deputy Minister Edward Omar Sharif Hiariej outlined one practical use: electronic interrogation reports assisted by AI. The goal is straightforward-reduce intimidation or violence during examinations by having statements typed automatically, with the examined party or suspect reviewing and signing the report.
What changes on the ground
- AI-assisted interrogation reports: Statements can be transcribed automatically, then verified and signed by the person examined.
- CCTV requirement: Under KUHAP Article 30, examinations of suspects must be recorded on surveillance cameras. These recordings may be used for investigation, prosecution, and defense.
- Legal timeline: KUHAP was signed on December 17, 2025 and took effect on January 2, 2026 (Article 369). The Draft Presidential Regulation is being prepared, with further technical details on recordings to follow in a Government Regulation.
"We are preparing all these advances for the implementation of our Criminal Procedure Code," the minister said.
Implications for ministries and agencies
- Update SOPs and forms: Define AI use during examinations, verification steps, and how the final statement is signed (paper or digital, as regulated).
- Procurement criteria: Require tamper-evident logs, clear accuracy benchmarks, Bahasa Indonesia support, and secure deployment options.
- Data protection: Set access controls, encryption, retention schedules, and roles for system administrators and reviewers.
- Human oversight: Keep investigators responsible for accuracy; the person examined must review the generated text before signing.
- Interoperability: Plan data flow across police, prosecutors, defense, and courts to avoid format and system lock-in.
- Training and change management: Train examiners, supervisors, and IT staff on process, ethics, and evidence handling with AI in the loop.
Safeguards to build in from day one
- Evidence integrity: Use time-stamps, hashing, and audit trails for both AI-generated reports and CCTV files.
- Fairness and accuracy: Monitor transcription quality, handle dialects and accents, and document known limitations.
- Rights information: Standardize how rights are communicated and logged during AI-assisted examinations.
- Audit access: Ensure timely access to recordings and logs for investigators, prosecutors, defense, and authorized oversight bodies.
- Fallback procedures: Define what happens if AI tools fail, are unavailable, or produce low-quality output.
Immediate next steps for government leaders
- Appoint an implementation lead and cross-agency working group.
- Run short pilots in a few jurisdictions; measure accuracy, time savings, user satisfaction, and complaint rates.
- Draft SOPs, consent language, and checklists aligned with the coming Presidential Regulation and Government Regulation.
- Budget for secure infrastructure, training, and independent testing.
- Set a communications plan for internal teams and the public to build trust.
Reference and capacity-building
- OECD AI Principles offer practical guardrails on safety, accountability, and transparency that can inform rollout.
- For teams planning skills upgrades on AI policy and operations, see AI course paths by job.
Your membership also unlocks:
