AI governance has no owner at all in almost one in five legal organisations, according to a LexisNexis survey of 543 UK lawyers and legal professionals reported on 14 October 2026. The report adds that the share is higher in small firms, at the Bar and in academic institutions.
I read that as the most useful number in the report for a firm of your size. A duty that nobody owns tends to be done by whoever happens to notice it, and AI fits that pattern closely. Three quarters of respondents use AI at least once a week and more than a third use it every day, yet only 29% say it is embedded in their organisation’s strategy and operations.
Read the figures with their source in mind. LexisNexis sells AI research tools, and its July 2026 sample mixes private practice, at 43%, with in-house lawyers, the Bar, the public sector and academia, so it is not a survey of small firms. The report gives no separate figure for small firms, only that the share with no owner is higher there.
Why a policy on its own falls short
The report argues that a policy reviewed once or twice a year cannot keep up with tools that change every month, and that governance has to work as a loop. A lawyer spots a fault, someone with authority assesses it, the guidance changes and the team hears about it. Concern about inaccurate or fabricated AI output has risen from 57% in early 2024 to 83% this year, so people are seeing the faults. What most firms lack is the second step in that loop, a named person who receives the report and has the authority to change the rule.
Where ownership exists it is often split. Some 39% of respondents named several teams as responsible, with IT the most common at 38% and risk and compliance next at 34%. A large firm can make a committee work. In a firm of twenty people, a duty shared between the practice manager, the IT contractor and the COLP belongs to none of them when a fee earner finds a fabricated citation at six in the evening.
Naming the owner in a small firm
The owner does not need to be a technologist. It needs to be someone senior enough to stop a tool being used and close enough to the work to hear when it goes wrong, which in most small firms means a partner or the COLP. Put that name in your AI use policy, beside the route by which anyone in the firm reports a fault and a promise that every report gets an answer. The SRA’s proposed changes to the statement of solicitor competence point the same way, because supervision and AI both feature in them.
Then give the loop a rhythm. A standing item at the monthly partners’ meeting, covering what went wrong, what changed and which tools staff have started using without approval, does more than an annual review of a long document. The last of those questions matters most, since the tools nobody approved are the ones no owner can govern.
Start this week by writing one name against AI governance and telling the whole firm whose name it is.
The LexisNexis report, Why AI governance needs a feedback loop, is on lexisnexis.co.uk with its methodology at the foot of the page, and Legal Futures reported its findings on 14 October 2026.
If your firm has not yet settled who owns AI, or how a fault reaches that person, start with a conversation.
