In short: The BSB has not banned barristers from using artificial intelligence. Its guidance, published on 18 May 2026, creates no new rules. It explains how the existing Core Duties and BSB Handbook requirements already apply when AI is used. A barrister remains personally responsible for the accuracy of everything submitted in their name, and cannot transfer that responsibility to a software provider. Where fabricated citations reach a court, referral to the regulator is the likely consequence.
Artificial intelligence is becoming an increasingly common tool within legal practice. From legal research and document preparation to administrative support, AI has the potential to improve efficiency at the Bar. However, the increasing use of artificial intelligence in legal practice has prompted courts and regulators to consider the risks associated with inaccurate AI-generated material, confidentiality, and the continuing professional responsibilities of lawyers.
Recent judicial decisions and regulatory developments have made clear that AI does not alter a barrister’s existing professional obligations. The central principle remains straightforward: a barrister is personally responsible for the accuracy and propriety of all work submitted in their name, whether or not AI has assisted in its preparation.
What did the Ayinde judgment say about AI in court?
The issue came into sharp focus in R (Ayinde) v London Borough of Haringey and Al-Haroun v Qatar National Bank [2025] EWHC 1383 (Admin), a Divisional Court judgment handed down on 6 June 2025 by Dame Victoria Sharp, President of the King’s Bench Division, and Mr Justice Johnson.
In one of the cases before the court, judicial review grounds prepared by a barrister contained five authorities that did not exist. When copies were requested by the opposing party, none could be produced. The court found that the threshold for contempt proceedings had been met, although contempt proceedings were not ultimately pursued. Wasted costs orders were made, and the barrister was referred to the Bar Standards Board.
The judgment did not create new professional duties. Instead, it reinforced long-established obligations requiring legal representatives to verify authorities, ensure accuracy in court documents, and avoid misleading the court. The judges also highlighted the importance of professional supervision and encouraged legal institutions to consider the implications of AI misuse within legal practice.
The court’s assessment of the technology was blunt. It stated that generative AI tools are not capable of conducting reliable legal research, and that answers which appear coherent and plausible may be entirely incorrect, citing authorities that do not exist or misstating those that do.
What is the Hamid jurisdiction?
Both cases were heard together under the court’s Hamid jurisdiction, named after R (Hamid) v Secretary of State for the Home Department. It is the court’s inherent power to regulate its own procedures and to enforce the duties that lawyers owe to the court.
The route the Ayinde matter took is worth understanding, because it explains why there are two citations in circulation. The wasted costs application was heard first by Ritchie J, reported at [2025] EWHC 1040 (Admin), who then referred the matter to the Hamid judge for further consideration. The Divisional Court judgment at [2025] EWHC 1383 (Admin) is the one that contains the guidance for the profession.
The significance of the jurisdiction is that it sits outside the regulatory system. It allows the court itself to act on conduct that undermines the administration of justice, independently of anything the BSB or SRA may later do.
What were the actual consequences?
The outcomes in Ayinde were financial and regulatory rather than contemptuous. Wasted costs of £2,000 were ordered against the barrister, with a corresponding order against the law centre instructing her, and a further sum of the law centre’s own fees was written off. Both the barrister and the solicitors were required to report themselves to their respective regulators, and the court made its own referral in addition.
The operative principle. The judgment states that where false citations are placed before the court, whether through a lack of proper checks or otherwise, referral to a professional regulator is likely to be appropriate. That is the sentence practitioners should take from the case.
Has it happened again since?
Yes, and that is the more troubling part of the picture. The Upper Tribunal (Immigration and Asylum Chamber) has since held its own Hamid hearing in a case involving fabricated citations, guiding itself by Ayinde and emphasising that the primary duty of solicitors and barristers is to the court and to the cause of truth and justice.
The pattern matters because it undercuts the assumption that Ayinde was an isolated lapse by one inexperienced practitioner. The courts have treated the publication of guidance as insufficient on its own, and have indicated that steps are needed to ensure guidance is actually followed.
What does the BSB’s AI guidance require of barristers?
Against this background, the Bar Standards Board published its Guidance on the Use of Artificial Intelligence and Other Technologies on 18 May 2026.
The BSB made clear that the guidance does not create new rules. Instead, it explains how the existing Core Duties and requirements in the BSB Handbook apply when barristers use AI and other technologies.
How does the duty to the court apply to AI-assisted work?
The BSB emphasises that the duty to the court remains paramount. Barristers must carefully review and verify any AI-assisted work before relying on it in pleadings, written submissions or oral advocacy.
The use of AI does not transfer responsibility to software providers or technology platforms. If inaccurate information is presented to a court, responsibility remains with the barrister who submits it.
Can a barrister blame AI for inaccurate material?
The guidance highlights that barristers must maintain the highest standards of honesty and integrity when using AI.
A barrister cannot excuse inaccurate, misleading or fabricated material by claiming that it was generated by an AI system. Professional accountability remains with the individual practitioner.
Ayinde illustrates how the court responds to attempts to minimise the problem after the event. Describing fabricated authorities as citation errors, or as cosmetic, was treated by the court as aggravating rather than mitigating. The response to discovery of an error is itself conduct the court assesses.
What are the confidentiality and data protection risks?
The BSB warns that the use of AI tools may create risks relating to confidentiality, legal professional privilege and data protection.
Before entering client information into any AI system, barristers should understand how that information is processed, stored and used. They must ensure that their use of technology complies with their duties of confidentiality and relevant data protection obligations.
The practical questions are specific. Does the provider use submitted material to train its models? Is data retained, and for how long? Where is it stored, and who within the provider can access it? A general-purpose consumer tool and a legal platform offered under a professional contract may give very different answers, and the answer determines whether client material can properly be entered at all.
What does technological competence mean in practice?
The guidance also emphasises the importance of technological competence. Barristers should understand both the capabilities and limitations of the AI tools they choose to use.
This includes recognising the risk that AI systems may generate inaccurate information, sometimes described as “hallucinations”, including incorrect case citations, quotations or legal analysis.
Competence here is not a technical skill. It is knowing what a tool cannot do, and in particular that fluent, confident output carries no indication of accuracy. The most dangerous fabrications are the plausible ones: a citation in the correct format, attributed to the right court, in the right year, for a proposition that sounds entirely orthodox.
Must a barrister tell the court that AI was used?
Not at present, provided it was used responsibly. The judiciary’s guidance for judicial office holders states that legal representatives are responsible for the material they put before the court and are under a professional obligation to ensure it is accurate and appropriate, but it does not currently impose a requirement on representatives to disclose their use of AI in preparing submissions.
That is the position rather than a guarantee. The same guidance contemplates that a judge may ask what checks were carried out to verify information obtained through AI, which means a practitioner should be able to answer that question about any document bearing their name.
What does the judiciary’s own guidance say?
The judiciary published updated AI guidance for judicial office holders in October 2025, replacing a version issued in April of that year. It applies not only to judges but to tribunal members, clerks, judicial assistants, legal advisers and support staff.
On fabrication the guidance is direct: AI tools may invent cases, citations and quotations, and may refer to legislation, articles or legal texts that do not exist. For practitioners, the relevance is that the bench is now explicitly alert to the problem, and the question of what verification was undertaken is one judges have been told they may ask.
What does the Bar Council’s guidance add?
The Bar Council has also issued guidance on the responsible use of generative AI. Unlike BSB guidance, it does not have the same regulatory status under the BSB Handbook, but it provides practical assistance to barristers using AI tools.
The Bar Council’s guidance discusses both general-purpose AI systems and legal-specific platforms. It emphasises that AI may assist legal work, but its outputs must be independently checked and lawyers must continue to apply their own professional judgment.
When launching the updated guidance, Barbara Mills KC, then Chair of the Bar Council, warned that recent case law had highlighted the dangers of AI misuse by lawyers and the potential consequences for public confidence in the administration of justice.
What does an adequate verification process look like?
Neither regulator prescribes a method, but the case law indicates what the courts expect. The Upper Tribunal has described checking the veracity of AI-produced citations against reputable legal search engines as the baseline a practitioner must meet before relying on them.
In practice that means a short, non-negotiable sequence for any authority that originates from an AI tool:
- Confirm the case exists, against a reliable source such as Find Case Law, run by The National Archives with the Ministry of Justice, HMCTS and the Judicial Office, or an established subscription service.
- Confirm the neutral citation is the one allocated by the court, and that the court, division and year match.
- Open the judgment and read the paragraph relied on, rather than the summary of it.
- Confirm the case remains good law and is not under appeal or overtaken.
- Check any quotation word for word against the judgment, including the paragraph number.
The discipline is the same one that applies to any authority found by any means. What AI changes is the base rate: a citation found in a textbook is almost certainly real, while a citation produced by a general-purpose model may not be, and the output gives no signal either way.
What does this mean for chambers and supervision?
The Ayinde judgment drew attention to supervision, and that is where the implications for chambers sit.
Pupils and junior tenants are the most likely to use these tools for research and the least likely to have the experience to recognise a plausible fabrication. A pupil supervisor who reviews a note for legal reasoning without checking whether the authorities exist is not performing the check the court now expects of the set.
For chambers the practical questions are whether there is a stated position on which tools may be used and for what, whether client-confidential material may be entered into any of them, how AI-assisted research is verified before it reaches a client or a court, and who is accountable for that verification. A set without answers is relying on individual practice, which is precisely the gap the recent cases have exposed.
What happens after a referral to the BSB?
A referral is the beginning of a process, not a finding, and the distinction matters.
Complaints and referrals are assessed by the BSB’s Investigations and Enforcement team. Depending on the seriousness of the matter, it may be concluded by the BSB itself, dealt with under the Determination by Consent procedure where the barrister agrees and does not dispute what happened, considered by a panel of the Independent Decision-making Body, or referred to a Disciplinary Tribunal convened by the Bar Tribunals and Adjudication Service.
The functions are separate. The BSB brings charges of professional misconduct; BTAS panels adjudicate them and decide sanction. An administrative sanction imposed by the BSB is not a disciplinary sanction and does not give the barrister a disciplinary record. A referral, an investigation or a listed hearing is not a finding of misconduct, and a decision that is under appeal is not final.
Does AI change a barrister’s professional judgment?
Neither the BSB nor the Bar Council has prohibited the use of artificial intelligence by barristers.
The regulatory approach is that AI can provide benefits when used appropriately, but it does not reduce a barrister’s responsibilities to the court, clients or the wider administration of justice.
The key lesson from the recent guidance and case law is that technology does not replace professional judgment. Barristers must continue to verify authorities, check the accuracy of legal analysis, protect confidential information and ensure that all work carried out in their name meets the standards required by the BSB Handbook.
What is the Civil Justice Council consulting on?
The wider justice system is also considering whether procedural changes are needed to address the increasing use of AI in litigation.
In February 2026, the Civil Justice Council launched a consultation examining the use of AI in the preparation of court documents. The proposals considered different categories of documents, including witness statements and expert reports.
One proposal was that legal representatives could be required to confirm that AI had not been used to generate or alter the substantive content of witness evidence. The consultation has concluded and the Civil Justice Council’s final recommendations are awaited.
If a confirmation requirement is adopted, the position described above on disclosure would change for the documents covered, and practitioners should watch for the final recommendations.
Key takeaways
- The BSB guidance, published 18 May 2026, creates no new rules. It applies the existing Core Duties and Handbook requirements to the use of AI.
- Responsibility cannot be transferred to a software provider. The barrister who submits the work answers for it.
- In Ayinde [2025] EWHC 1383 (Admin), judicial review grounds contained five authorities that did not exist. Wasted costs of £2,000 were ordered against the barrister, who was referred to the BSB.
- Where false citations reach a court, the judgment indicates that referral to a regulator is the likely consequence.
- The Hamid jurisdiction allows the court to act on duties owed to it, independently of anything the regulator does.
- There is currently no requirement to disclose AI use to the court, but a judge may ask what verification was carried out.
- Bar Council guidance is practical assistance, not regulation, and does not carry the same status under the Handbook as BSB guidance.
- A referral is not a finding. The BSB brings charges; BTAS tribunals adjudicate them and decide sanction.
Frequently asked questions
Can barristers use AI?
Yes. Neither the Bar Standards Board nor the Bar Council has prohibited the use of artificial intelligence. The BSB’s position is that AI can offer benefits when used appropriately, provided the barrister’s existing professional obligations are met in full.
Does the BSB’s AI guidance create new rules?
No. The BSB has made clear that the guidance, published on 18 May 2026, creates no new rules. It explains how the existing Core Duties and requirements in the BSB Handbook apply when barristers use AI and other technologies.
Who is responsible if AI produces an inaccurate citation?
The barrister who submits the work. The use of AI does not transfer responsibility to software providers or technology platforms, and a barrister cannot excuse inaccurate or fabricated material on the basis that an AI system generated it.
What happened in the Ayinde case?
Judicial review grounds prepared by a barrister contained five authorities that did not exist, and none could be produced when requested. The Divisional Court found the threshold for contempt proceedings had been met, though contempt was not pursued. Wasted costs of £2,000 were ordered against the barrister, who was referred to the Bar Standards Board.
What is the Hamid jurisdiction?
The court’s inherent power to regulate its own procedures and enforce the duties lawyers owe to it, named after R (Hamid) v Secretary of State for the Home Department. It operates independently of the professional regulators, so a practitioner can face action from both.
Do barristers have to tell the court they used AI?
Not currently, provided the tool was used responsibly. The judiciary’s guidance does not require disclosure, but it does contemplate that a judge may ask what checks were carried out to verify AI-derived material.
Can barristers put client information into AI tools?
Only where doing so is consistent with their duties. The BSB warns that AI tools may create risks for confidentiality, legal professional privilege and data protection, and barristers should understand how information entered into a system is processed, stored and used before entering it.
Does Bar Council AI guidance carry the same weight as BSB guidance?
No. Bar Council guidance offers practical assistance but does not have the same regulatory status under the BSB Handbook as guidance issued by the regulator.
What is technological competence?
Understanding both the capabilities and the limitations of the AI tools in use, including the risk that a system will generate inaccurate material such as incorrect case citations, quotations or legal analysis.
What happens if a barrister is referred to the BSB over AI misuse?
The matter is assessed by the BSB’s Investigations and Enforcement team. It may be concluded by the BSB, dealt with by Determination by Consent, considered by the Independent Decision-making Body, or referred to a Disciplinary Tribunal convened by BTAS. A referral is not itself a finding of misconduct.