evidoria

← Back to browse

Good practice Imported

Oman's Public Prosecution AI Pilot — Machine-Assisted Review of Judicial Case Files

Oman · Muscat · See the Oman profile · See the Muscat profile

Evidence: Descriptive / self-reported Top 88% 33/100 · Ask Evidence Copilot about this practice

As part of a wider 2025 national AI rollout, Oman's Public Prosecution is piloting a tool built with local firm Rahal that reads case files, drafts questions and recommends dismissal or referral, reporting over 90% accuracy while keeping data processed locally.

>90% %
Reported accuracy rate

Details

Maturity
Pilot
Promoter
Oman Public Prosecution (with Rahal)
Period
2025
Keywords
justice, law enforcement, judiciary

Context

In 2025, Oman's Ministry of Transport, Communications and Information Technology oversaw a wave of government AI pilots spanning health, justice, public tenders, higher education and cybersecurity. Within that programme, Oman's Public Prosecution began testing an AI tool for reviewing judicial case files, built with local technology firm Rahal.

Activities

The system analyses case documents, generates structured questions, and drafts recommendations on whether a case should be dismissed or referred onward, processing sensitive data locally rather than sending it to external servers.

Results

Oman Observer and the SAMENA Telecommunications Council's industry news service, citing regional outlet MEA TechWatch, both reported the pilot achieved 'an accuracy rate exceeding 90 per cent.'

Conclusions

Neither source explains what the accuracy percentage measures, nor describes an appeals process, bias testing, or how a prosecutor's use of an AI-drafted recommendation is checked — a meaningful gap given the tool influences decisions on whether a criminal case proceeds.

Implementation
Indicative cost
Low (< €50k) — Built with a local vendor (Rahal); no budget figure is published.
Time to results
Short (< 1 year) — Piloted within 2025, no reported timeline for wider judicial rollout.
Staffing & skills
Oman Public Prosecution, Rahal (local technology vendor)

Conditions for success

  • Data processed locally rather than sent to external servers

Common failure modes

  • No disclosed appeals process, bias testing, or human-review safeguard for AI-drafted case recommendations

Where it fits

Governance type
single government department pilot
Scale
single department
Income level
high income

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

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.

Similar practices you may find useful