Robes, Not AIs: Judges Get a Cautious Green Light on AI

AI is coming to the bench. On 9 October 2026, the senior judiciary of England and Wales published a Position Statement on the Judicial use of Artificial Intelligence, signed by Sir Colin Birss, President of the Business and Property Division and Lead Judge for Artificial Intelligence. The statement sets out how AI is used by judges, what safeguards apply and how the approach will be kept under review. It sits alongside the judiciary's AI Guidance, which was first issued in December 2023 and updated in April and October 2025.

The central message is that AI "will be adopted to support but not replace judicial decision-making" and will be "subject to strict safeguards and robust testing". Judicial decisions will continue to be made by a judicial office holder. AI may assist, but it must not displace judicial reasoning, the evaluation of evidence or the exercise of judicial discretion. The statement says that "human oversight is essential". Every AI output used in judicial work must be reviewed, verified and approved by the responsible judge, who remains fully accountable for it.

The statement follows pilots that focused on lower-risk tasks. These included summarisation, transcription, redaction, chronology-building and document analysis. Only approved tools may be used for judicial work. Those tools must keep data in the UK, have enterprise-level data protection and have web access disabled. A centralised steering group, chaired by the Lead AI Judge, will consider any new use cases, products or pilots. It will assess the legal, ethical, operational, security and data protection issues before anything can go ahead. The Judicial College will provide training to support confident and appropriate use.

The following are key:

1. Argue to a human decision-maker. Outcomes, the evaluation of evidence and the exercise of discretion stay with the judge, so submissions in commercial and finance disputes should still be built around persuasive reasoning, not shortcuts.
2. Make document-heavy bundles easy to read. Pilots have tested AI for summarisation, chronology-building and document analysis, so clear, well-structured chronologies and bundles in complex banking and commercial cases are likely to pay off.
3. Protect confidential and sensitive material. Approved judicial tools must keep data in the UK, have enterprise data protection and have web access disabled, and the pilots looked closely at how sensitive material is handled. Clients with trade secrets, IP or market-sensitive financial data should keep confidentiality protections and redaction strategies under active review.
4. Hold your own AI use to the same standard. The judiciary insists that every AI output is reviewed, verified and approved by an accountable person, and its guidance highlights the risks of hallucination and bias. Firms should check all AI-assisted citations, authorities and analysis in the same way before filing.
5. Track the framework as it develops. The steering group will revise approved uses, safeguards and governance as pilots, incidents and risks emerge, so practitioners should watch for updates alongside the judicial AI Guidance.

The machines may help with the paperwork, but the judgement will remain human.