AI in law

The SRA says firms remain responsible and accountable for the outputs from AI they use. If a third-party chatbot gives initial advice, the firm is responsible for its errors, and clients should be suitably informed of how AI is involved in their matters.

The SRA also warns against trusting a system to judge its own accuracy and calls for supervision that can cope with the speed AI brings.

When it authorised Garfield.Law, the SRA checked quality control, confidentiality and conflicts, and noted that the system will not propose relevant case law because that is a high-risk area for large language models.

The Law Society's generative AI essentials says solicitors' duties to court and client apply whether AI is used personally or by someone they supervise, and that citations in final documents must be verified against reliable sources.

In Ayinde v London Borough of Haringey, the Divisional Court dealt with referrals arising from the actual or suspected use of generative AI to produce legal arguments or witness statements that were not checked, so that false information, typically fake citations or quotations, reached the court.

The court said the facts raised concerns about the competence and conduct of the individual lawyers, and wider concerns about the adequacy of training, supervision and regulation across the profession.

It was not an isolated case. The Law Society notes that in April 2026 a judge in Rodney v Gee'z Micro Bar & Pitstop referred two solicitors to the SRA after AI-assisted court documents contained inaccurate citations.

The lesson for AI in law is simple and practical: no citation, quotation or authority goes into a document until a person has checked it against a primary source.

The Legal Services Board's June 2026 research says AI-powered legal tools could make a real difference for people who lack access to legal help, and that around 32% of adults in England and Wales who experience a legal problem receive no professional support.

In its consumer survey, 70% expected AI to improve ease of access to legal services and 64% expected greater affordability. Respondents also named five protections they consider non-negotiable: a minimum guarantee for accuracy, no consequential action without informed consent, some human oversight, access to redress and protection of personal data.

The LSB has also joined the government's AI Growth Lab, an advisory sandbox where providers test AI products under regulatory supervision. Participation does not constitute regulatory approval.

For firms, AI in law is therefore partly about efficiency and partly about serving clients who previously could not afford a lawyer at all.

A baseline is useful even if you decide to keep the process manual. Measure the work that matters, then compare quality and review effort as well as time.

Assess the workflow before changing it, or plan business AI training if the gap is how people use existing tools.