AI Governance in Criminal Justice Gains Focus With New Standards and Research

Efforts to establish governance for artificial intelligence in criminal justice have accelerated, with multiple organizations releasing standards and research findings in 2025 and 2026. These developments address both the opportunities and challenges presented by AI tools in legal processes.

The Council on Criminal Justice formed a Task Force on Artificial Intelligence in June 2025, aiming to create standards and recommendations for the responsible use of AI in the criminal justice system. The Task Force published its initial set of five guiding principles in October 2025, focusing on safe, ethical, and effective AI deployment.

Stanford Law School took further steps by appointing a former California State Senate AI adviser in February 2026 to lead its AI Initiative. This initiative is designed to advance legal frameworks, policy guidance, and training related to AI systems in law.

Research published in March 2026 examined the influence of AI decision-aid tools on pretrial and sentencing decisions. The review found that the observed effects of these tools were limited or absent, and it also noted a lack of comprehensive empirical studies in this area.

What the numbers show

• The ExpungeMate project increased expungements at a single event from 70 to 751
• Citizens Advice’s chatbot “Caddy” halved advisor response times during a six-week trial
• Google’s Notebook LM tool can process up to 1 million tokens of information

At the second annual AI & Access to Justice Summit in November 2025, the Stanford Legal Design Lab highlighted projects using AI to improve legal aid. The ExpungeMate project, for example, automated expungement petitions and substantially increased the number processed at a single event. During the same summit, Citizens Advice reported that its internal chatbot “Caddy” reduced advisor response times by half in a six-week test.

Other technology platforms have introduced AI features to support legal work. Relativity’s e-discovery platform, through its “Justice for Change” program, provided generative AI tools for document review and translation in over 100 languages. Google’s Notebook LM tool was noted for its ability to analyze large volumes of information with an expanded context window.

Despite these advancements, concerns remain about the risks associated with generative legal AI systems. A March 2026 study identified issues such as hallucination and overreliance, which can affect explainability and present challenges to judicial independence and fundamental rights.

The article “AI in Criminal Justice: Why Governance Matters and.

How to Make It Work” was published in March 2026, summarizing recent progress and ongoing challenges in the governance of AI within the criminal justice sector. These developments reflect ongoing efforts to balance innovation with safeguards in legal technology.

* This article is based on publicly available information at the time of writing.