---
title: "Justice AI Unit"
category: "companies"
aliases: ["Justice AI", "MoJ Justice AI Unit"]
website: "https://ai.justice.gov.uk/"
linkedin: "https://www.linkedin.com/company/justice-ai-unit/"
sourceLabels: ["Justice AI Unit official site", "No10 Innovation Fellowship", "YouTube transcript", "Synthesis"]
highlighted: "true"
highlightPriority: "high"
---
# Justice AI Unit

## Overview
The Justice AI Unit is a dedicated AI team within the UK [[ministry-of-justice|Ministry of Justice]]. Its public site describes it as an interdisciplinary team of AI specialists, designers, technologists, and operational experts working to embed responsible AI across the justice system.

## Conference Relevance
The unit is the central organization in [[youtube-qlHaO6laBlM|Shipping Production AI Inside Government]]. It demonstrates a conference-relevant model for production AI: small, embedded teams working directly with users, shipping tools into real operational environments, and balancing speed with public-sector trust.

## Connections
- [[william-tarr]]
- [[louis-allgood]]

## Sources
- [Justice AI Unit official site](https://ai.justice.gov.uk/)
- [No10 Innovation Fellowship: Transforming Justice with Responsible AI](https://fellows.ai.gov.uk/articles/william-tarr-moj/)
- [No10 Innovation Fellowship roster](https://fellows.ai.gov.uk/)

## Evidence Graph
This evidence graph summarizes how this organization appears across the conference source graph: scheduled sessions, linked videos, transcripts, and slide-derived evidence.

### Linked Sessions
- No related scheduled session was found in this pass.

### Media Signals
No linked video, transcript, or slide source has been attached yet.

## Evidence Boundary
The Justice AI Unit official site supports the organizational description. The highlighted YouTube transcript supplies the conference-specific synthesis and should be treated as auto-captured source text.

## Operating Model Signals
- Work directly with users rather than only through headquarters requirements.
- Pair engineers with operational experts who know prisons, probation, courts, and internal justice workflows.
- Use small-product loops: identify pain, build, test, feature-flag, scale, or kill.
- Design for the actual deployment environment, including offline and low-connectivity situations.

## Related Source Pages
- [[youtube-qlHaO6laBlM]]
- [[youtube-qlHaO6laBlM-slides]]
- [[youtube-qlHaO6laBlM-transcript]]