On 18 May 2026, the BSB published its Guidance on the use of Artificial Intelligence and Other Technologies (the BSB Guidance). The BSB Guidance follows hot on the heels of guidance published by the Law Society, the SRA, RICS and FRC each of which has been considered on this hub.
Jamie Smith KC, Clare Dixon KC and Seohyung Kim consider the BSB’s Guidance.
Three Key Points
First, the BSB supports AI. It encourages adoption of technologies, including AI, where they improve practice management or the quality and availability of legal services, so long as it is done competently.
Second, the BSB Guidance is broad. It is not confined to generative AI but also address agentic AI and, more broadly, “other technologies.” That matters because the direction of travel is increasingly towards autonomous tools, which are likely to raise more difficult regulatory problems.
Third, the BSB Guidance does not create new duties. Rather it seeks to explain how the existing Core Duties and rules are likely to be applied in a working environment using AI. AI may change the mode of working, but the allocation of responsibility remains the same.
BSB’s basic position: use AI but beware of serious consequences
Much of the BSB Guidance will feel unsurprising. Barristers must not mislead the court or the client; must protect confidentiality; must act competently; and must manage practice risks properly. All of those duties are engaged by AI use. In particular, presenting inaccurate AI-material to the court may amount not merely to negligence but to recklessness.
This is sharp warning. The BSB Guidance treats AI error as an entirely foreseeable professional risk, with potentially serious consequences if the barrister fails to verify what is being put before the court.
In order to mitigate the risk, barristers are encouraged to review the potential risk or harm that can result from the use of AI, which is illustrated in a table identifying areas of low, medium and high risk. A prudent and appropriately defensive approach would be to maintain a record where the work done falls into the “likely high risk” category.
Competent use of technology, including by others
One of the more striking aspects of the BSB Guidance is that technological competence is not limited to tools a barrister personally chooses to use. The BSB expects barristers to maintain a sufficient level of competence to understand the effects of AI used by clients, solicitors, courts and opponents. In other words, when it comes to AI, the “ostrich” defence will not do.
Delegation and individual responsibility
The BSB Guidance raises an important question: how far may an individual barrister rely on others when deciding whether an AI product is compliant and safe?
The BSB Guidance acknowledges that decisions about whether a particular AI tool are compatible with BSB regulations and an individual barrister’s legal obligations may well be taken at a Chambers level. It also treats AI use as a species of outsourcing and emphasises the need for appropriate policies, data governance and confidentiality controls.
But there is a real tension. A hallucinated authority is easy enough to check. Confidentiality and data handling are different: if a vendor says prompts are secure or that data is not used for training, how could an individual barrister verify that independently? In most cases they cannot. And the ramifications of a data leak may not emerge until much later.
It remains to be seen the extent to which individual barristers will be able to rely upon the work of Chambers’ committees when it comes to the institutionalised security and confidentiality features of an AI product (and the adequacy of Chambers’ Policies in that regard).
We anticipate that any regulator would not allow that reliance to be slavish and it would remain an individual barrister’s responsibility to sense check the output of specialist Committees within their Chambers and raise points of concern if they have any. Collective governance can assist individual compliance but it does not replace individual responsibility.
Use particular caution with Agentic AI
The BSB Guidance says agentic AI (i.e. systems capable of coordinating and executing multiple tasks with some degree of autonomy) are likely to be high risk for most applications at the Bar and should be treated with absolute caution.
In taking actions or progressing tasks with a degree of autonomy beyond merely generating text in response to a manual prompt, Agentic AI introduces a further difficulty. In this situation, the gap between what the barrister thinks the system is doing and what it is actually doing may become professionally dangerous.
A practical document
The BSB Guidance is practical. Its emphasis on risk categorisation, secure use, record-keeping, governance policies and verification procedures nudge Chambers and practitioners towards operational competence.
In that respect, the BSB Guidance is part of a broader shift: regulators are moving away from asking whether AI should be used, and towards asking how a professional can demonstrate that AI was used responsibly, proportionately, and defensibly.
Conclusion
The BSB Guidance is measured and realistic. It supports the use of AI and existing professional rules remain relevant in assessing responsible use of AI. Its central message is simple: barristers may use AI but they may not offload judgement or responsibility to it.
© Jamie Smith KC, Clare Dixon KC and Seohyung Kim
This article is not intended as a substitute for legal advice. Advice about a given set of facts should always be taken.
