Law Society AI Practice Note: Responsibility Stays With the Solicitor

The Law Society’s message is simple: if AI gets it wrong, it’s still on you. Its practice note on using AI in legal practice, published October 21, 2026 and running to 58 pages, the Law Gazette reports, says professional responsibility for AI output stays with the solicitor. It also says something many firms would rather not say out loud: leaning on AI could leave junior lawyers without the skills they’ll need to supervise it.

As Artificial Lawyer pointed out, the Law Society is a representative body rather than the regulator, so the note carries no legal authority on its own; the Solicitors Regulation Authority regulates solicitors in England and Wales.

What Does the Practice Note Ask Solicitors to Do?

The note says there is no AI-specific regulator for legal practice and that existing duties apply, including the SRA’s Standards and Regulations. It says the SRA does not authorise, certify or approve AI systems, and that self-labelling by AI or legal tech companies does not guarantee a system is appropriate or safe for legal practice. It also points to an SRA warning notice on AI misuse issued in August 2026, which flagged hallucinated citations and client data put into tools without proper safeguards.

Law Society president Brett Dixon summed it up:

“Professional responsibility cannot be transferred to any AI system. Solicitors and law firms should choose their AI tools carefully and always check that legal work meets their professional standards.”

A few points go further than the usual guidance. Paying for a tool does not end the scrutiny: the note says an enterprise label alone does not guarantee enough protection, and that firms should check contracts, privacy terms and technical settings, including where data is stored and processed. Self-checking AI does not count either. The note says an AI system “cannot reliably validate the accuracy of its results on its own” and that relying on AI to verify AI outputs will not satisfy professional obligations if there are inaccuracies.

The billing line is the one clients will quote back to you: where fees are time-based, don’t bill for the hours a task would have taken without AI. Time spent operating the tool, reviewing its output and applying judgment may be charged where the client care letter allows. The note also encourages transparency about AI use in client care letters or terms of engagement, and says solicitors may ask clients to disclose AI-generated material they send in.

Who Will Train the Lawyers Who Check the AI?

This is the most human part of the note. It warns that overreliance on AI may erode skills, and says this risk may be particularly significant for junior and aspiring solicitors. It asks supervisors to make sure AI does not replace opportunities for juniors to build research, drafting and analytical skills, and to learn to spot errors in AI output. Dixon went further, warning that overreliance “may result in a catastrophic gap of future legal skills and knowledge.”

Artificial Lawyer pushed back, arguing that some junior work can be replaced by AI with the right checks, and that the answer to the training problem is more AI, such as simulators.

We think both sides have a point, and it lands at a tense moment for people working in firms. CMS has begun preparing a formal consultation with its UK business services staff, with potential redundancies being explored alongside an initiative that sources said will look at how AI can be implemented, RollOnFriday reported on October 2, 2026. In May, A&O Shearman cut an undisclosed number of business-team staff after significant investment in technology and data, the Law Gazette reported. If you are a trainee or a paralegal reading this with some unease, that is a fair reaction. If junior work shrinks too, someone has to answer the obvious question: who trains the people who will check the AI in five years?

Why Should Lawyers Outside England and Wales Read It?

U.S. teams should read the billing section even though it does not bind them. It is the clearest statement we have seen from a major professional body on how hourly billing and AI efficiency should interact, and clients will quote it. Multinational legal departments that instruct UK firms should expect it to shape engagement letters.

Large firms with UK offices can check their AI policies against the governance section, which covers registers of approved tools, supervision checkpoints and audit; most of it maps to work these firms already do. In-house teams can borrow the vendor questions directly: where data is processed, who at the provider can see it, and how terms change after signing. Small firms may get the most practical value, because the note lays out risk by type of access, from free consumer tools to enterprise and internally built systems, in plain language. Just don’t go looking for a list of approved tools. The note says plainly that no one approves them.

The body that can enforce is the SRA, so its next move matters most. The Civil Justice Council, the note says, expects to publish its final report on AI in court documents later in 2026.

FAQ

Is the Law Society practice note binding?

No. As Artificial Lawyer pointed out on October 21, 2026, the Law Society is a representative body; the Solicitors Regulation Authority regulates solicitors.

What does the note say about billing?

Where fees are time-based, firms should not bill for the time a task would have taken without AI.

Should solicitors tell clients about AI use?

The note encourages transparency about AI use in client care letters or terms of engagement, per the Law Gazette’s October 2026 report.

What comes next?

The Civil Justice Council expects to publish its final report on AI in court documents later in 2026, the note says.

*Updated Oct 23 to refresh the story and add an FAQ.*