evidoria

← Back to browse

Good practice Imported

AI-Assisted Wealth-Report Screening — Indonesia's KPK Uses Machine Learning to Flag Anomalous Asset Declarations

Indonesia · Jakarta · See the Indonesia profile · See the Jakarta profile

Evidence: Descriptive / self-reported Top 98% 20/100 · Ask Evidence Copilot about this practice

Indonesia's KPK piloted AI that flags anomalies in officials' mandatory wealth declarations for manual review. In 2025 it screened about 1,000 of 415,062 required filings - under 0.3% - with KPK's chair citing efficiency gains, though it remains a small early pilot.

415,062
Officials required to file wealth declarations (2025)
~1,000
Reports screened via the AI pilot (2025)
under 0.3%
Share of total filings screened by AI (2025)
AI-Assisted Wealth-Report Screening — Indonesia's KPK Uses Machine Learning to Flag Anomalous Asset Declarations

Details

Maturity
Pilot
Promoter
Komisi Pemberantasan Korupsi (KPK) — Indonesia's Corruption Eradication Commission
Period
2025 pilot (publicly announced January 2026)
Keywords
anti-corruption, public integrity, asset declaration, AI governance

Context

Indonesia's Corruption Eradication Commission (KPK) requires public officials to file annual wealth declarations (Laporan Harta Kekayaan Pejabat Negara, or LHKPN) under Law No. 28 of 1999. Reports that pass an automatic check receive an immediate confirmation receipt; from 2025 the KPK began routing a subset of filings through an AI system that flags statistical outliers - asset values inconsistent with an official's declared income - for deeper manual review by the LHKPN team, who then cross-check banking and institutional records.

Results

Of the 415,062 officials required to file a declaration in 2025, KPK ran the AI screening on around 1,000 reports during this pilot phase - under 0.3% of total filers. KPK Chairman Setyo Budiyanto said the AI-verified reports in 2025 showed 'improved optimisation and efficiency,' with the system announced publicly in late January 2026.

Conclusions

This is a genuine, government-run deployment with a named official source and a specific coverage figure, but it is still a small-scale pilot covering a tiny fraction of total filers, with no published data yet on how many flagged cases led to further investigation or sanctions. It is a system worth tracking as it scales, not yet an evaluated success.

Implementation
Indicative cost
Low (< €50k)
Time to results
Short (< 1 year)
Staffing & skills
KPK LHKPN team

Conditions for success

  • existing legal framework (Law No. 28 of 1999)
  • named oversight team for manual cross-checking
  • banking/institutional record access

Common failure modes

  • no published data on how many flagged cases led to investigation or sanctions
  • AI model, criteria and error rate not disclosed
  • pilot covers under 0.3% of required filers

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful