Lawyer struck off in first disciplinary case on AI misuse

The Solicitors Disciplinary Tribunal (the Tribunal) has, for the first time, sanctioned a lawyer in relation to the misuse of generative AI in proceedings. In Solicitors Regulation Authority Ltd v Kumar (Case No. 12884-2026), decided on 25 August 2026, the Tribunal ordered that a Registered Foreign Lawyer (the Respondent) be struck off the Register of Foreign Lawyers. The case arose from two distinct sets of misconduct: first, a criminal conviction for knowingly employing a person without the right to work; and second, and of wider professional significance, the Respondent’s use of unverified and inaccurate legal citations arising from his use of generative AI in the course of the same disciplinary proceedings. Notably, the Tribunal was clear that the sanction it reached was the same sanction it would have imposed if either allegation had appeared before the Tribunal without the other, meaning that the AI-related misconduct alone was sufficiently serious to warrant a strike-off order.

BACKGROUND

The Respondent had represented an individual referred to as "the Client" in respect of immigration matters whilst operating through a law firm. The Client's claim for asylum was withdrawn, and he was classed as having been an overstayer in the UK. 

The Respondent was aware that the Client's asylum claim had been withdrawn and that a further leave to remain application had been rejected, leaving the Client without the right to live or work in the UK.  Home Office investigators subsequently found that the Client was working at the Respondent's firm in an administrative capacity, being paid cash in hand.  This resulted in the Respondent's conviction for a criminal offence contrary to Section 21(1) and (2) of the Immigration, Asylum and Nationality Act 2006.

The SRA brought disciplinary proceedings before the Tribunal on two grounds: 

1. The Rule 12 Allegation: The SRA brought an allegation against the Respondent on the basis that he knowingly employed an adult who was disqualified from employment by reason of immigration status, resulting in his criminal conviction. In doing so, the SRA alleged that the Respondent breached Principles 1, 2 and 5 of the SRA Principles 2019.
2. The Rule 14 Allegations: In the course of responding to the Rule 12 Allegation, the SRA alleged the Respondent had filed a written Answer and sent emails containing misleading references to authorities as a result of his use of AI. The SRA submitted that the authorities the Respondent cited either did not exist at all; had incorrect citations; misquoted the principles set out in the authorities; did not contain the passages quoted from them; or did not support the propositions for which they were cited. In doing so, the SRA alleged that the Respondent breached Principles 1, 2 and 5 of the SRA Principles 2019 and Paragraph 1.4 and Paragraph 2.4 of the SRA Code of Conduct for Individuals (the Code).

RULE 12 ALLEGATION: THE IMMIGRATION CONVICTION

The Respondent did not dispute the fact of his conviction, but asserted that he had originally pleaded not guilty and only changed his plea because he was suffering from severe physical and mental health difficulties and felt incapable of enduring a full criminal trial. The Respondent also contended that he had fully served his criminal debt to society by completing the Community Order and submitted that "permanent regulatory exclusion" would amount to "an unfair double punishment". 

The Tribunal found the allegation proved in full. It was satisfied that the certificate of conviction was proof of the findings of fact upon which the conviction was based, and that the allegation was properly made out. The Tribunal noted that dishonesty was not an element of the relevant criminal offence. It found that the Respondent had failed to uphold the rule of law and the administration of justice, thereby jeopardising public trust and confidence in the solicitors' profession.
 
RULE 14 ALLEGATIONS: THE (MIS)USE OF AI

In the course of defending the Rule 12 Allegation, the Respondent filed and served an Answer. Subsequently, the SRA identified incorrect citations in the Answer and citations which did not reflect the legal propositions outlined. The SRA submitted that the authorities cited by the Respondent either did not exist, had incorrect citations, misquoted the principles set out in the authorities, did not contain the passages quoted from them, or did not support the propositions for which they were cited. 

Following the SRA's identification of these issues, the Respondent sent an email admitting that he had used AI, and purported to correct the errors.  The SRA submitted that this email was itself drafted using generative AI and contained further false material.  The Respondent accepted that the email had also been produced with AI and contained inaccuracies, and stated that he was unable to provide copies of the authorities on which he relied. The Respondent later explained that he had no intention to mislead but simply did not have the expertise to verify the AI output.

THE TRIBUNAL'S ANALYSIS OF THE AI ALLEGATIONS

The Tribunal acknowledged that this was the first time a lawyer's use of AI or large language models in legal proceedings had been litigated before it. It accepted the SRA's position that the issue was one of profound public importance, relating to the proper role of lawyers and the effect that AI could have on the administration of justice and the rule of law.  

The Tribunal noted that the Respondent accepted that his actions breached Principle 1 (upholding the law and the proper administration of justice). As to Principle 2 (upholding public trust and confidence in the profession), he deeply regretted that his “technological negligence created a risk of bringing the profession into disrepute” but denied that his actions breached public trust and confidence in the profession. Regarding Principle 5 (acting with integrity), he denied “a lack of subjective ethical integrity”. The Respondent denied breaching Paragraphs 1.4 and 2.4 of the Code, denying any intention to mislead the Tribunal or advance legal submissions known to be false.

The Tribunal drew on Ayinde v Haringey LBC [2025] EWHC 1383 (Admin) (discussed further here), which it deemed the leading judgment on "the actual or suspected use by lawyers of generative artificial intelligence tools to produce written legal arguments … which are not then checked, so that false information (typically a fake citation or quotation) is put before the court”. The court in Ayinde held that those who use AI to conduct legal research have “a professional duty to check the accuracy of such research by reference to authoritative sources”, including "the Government's database of legislation, the National Archives, the official Law Reports and the databases of reputable legal publishers”. The Tribunal noted that the court further held that there are serious implications for the administration of justice and public confidence in the justice system if AI is misused, engaging Principles 1 and 2, and that practical and effective measures must ensure that every individual providing legal services understands and complies with their professional and ethical obligations when using AI.

The Tribunal made two further points of wider application:

1. It confirmed that there is no distinction in principle between misleading material sent to a regulator, a disciplinary tribunal, or a court. The Tribunal fulfils an important public function. 
2. The issue is ultimately about a lawyer's duty in litigation and to the court. Whilst AI was, in this case, the “medium” through which the erroneous material was produced, the same issue might arise using common search engines or where work was produced by another person for whom the respondent had ultimate responsibility. 

The Tribunal also rejected the Respondent's argument that the obligations set out in Ayinde did not apply to him because he was a Registered Foreign Lawyer rather than a solicitor, and because he was not in practice at the time, noting that the Respondent was a regulated individual who was obliged to comply with the standards of his profession. 

Accordingly, the Tribunal found that the allegations were proved insofar as they breached Principles 1, 2 and 5, and Paragraph 1.4 of the Code, further finding that the alleged breach of Paragraph 2.4 of the Code was proved in part in relation to two of the cases which were relied upon for propositions that those authorities did not support, and therefore they amounted to unarguable submissions.

SANCTION

The Tribunal's assessment of culpability and harm was unsparing.

In relation to the Rule 12 Allegation, the Tribunal determined the Respondent's culpability to be very high. His motivation was economic advantage and his actions in offering the Client informal employment were planned and a breach of his position of trust. The Tribunal noted that the Respondent was an immigration specialist who understood that what he was doing was outside the law. The Tribunal also assessed the level of harm to be very high, and found that the Respondent exploited the Client and that this constituted abuse of a position of authority.

In relation to the Rule 14 Allegations, the Tribunal also determined culpability to be very high. The Tribunal stressed that a lawyer is always responsible for the accuracy of his or her output. While mistakes happen, the Tribunal found that the Respondent had embarked on the use of AI without giving any proper thought to the consequences. He had failed to take proper steps to ensure accuracy. The Tribunal was especially critical of the fact that his misconduct was repeated after errors had been identified and explained to him, finding that “Rather than tak[ing] active steps to rectify his mistakes, the Respondent repeated them”, and the Tribunal gave very substantial weight to that repetition. 

The Tribunal further assessed the harm as high, commenting that it is a “cardinal duty” of a solicitor not to mislead a court or tribunal. The fact that the Respondent was acting for himself rather than a client did not operate as a mitigating factor: he was seeking to be the direct beneficiary of his own misconduct, and the harm from abrogating his responsibilities as a regulated lawyer had a potentially unlimited impact.

In arriving at the appropriate sanction, the Tribunal also placed significant weight on the need to protect the reputation of the profession, citing Bolton v The Law Society [1994] 1 WLR 512 and noting that "an important purpose of a sanction is to maintain the reputation of the solicitors' profession". It found that the damage to the reputation of the profession caused by the Respondent's misconduct was "extremely significant", as the public would trust a solicitor to respect the rule of law and the administration of justice, and concluded that "it was paramount that the reputation of the profession be protected".

The Tribunal found that the repeated disregard for verifying submissions had serious implications for the administration of justice and public confidence in the legal system, and ultimately determined that the Respondent's conduct was so serious that it required a strike-off order. The Tribunal was clear that it would have imposed the same sanction if either allegation had appeared before the Tribunal without the other.
 
COMMENT

At the time, the High Court in Ayinde had invited the legal regulatory bodies to consider, as a matter of urgency, what further steps, beyond issuing guidance, needed to be taken to address the problem. Now, the Kumar decision provides the profession's first clear disciplinary benchmark for AI misuse in legal proceedings, just days after the SRA published its Warning Notice on AI misuse on 17 August 2026 (discussed here), which flagged the risks of AI hallucinations in submissions to courts and tribunals as one of two primary concerns. The Kumar case illustrates precisely those risks in practice, and demonstrates the seriousness of the professional consequences of such conduct.

Three points stand out:

• First, the Tribunal was clear that the issue is not specific to AI; the same professional obligations apply whether erroneous material is generated by a large language model, a common search engine, or another person for whose work the solicitor had ultimate responsibility. This was ultimately an issue of the lawyer’s duty in litigation and to the court.
• Second, the Tribunal placed particular emphasis on the repeated nature of the misconduct, which was treated as a significant aggravating factor. The Tribunal gave very substantial weight in its finding of very high culpability to the fact that, having been put on notice of AI errors, the Respondent repeated the mistake rather than taking steps to rectify it. This approach mirrors Cork v Smith [2026] EWHC 1199 (Ch), where the court was particularly critical of the continued reliance, in that case by a City law firm, on misleading AI-generated material after the judge had alerted the firm to potential errors, as well as the firm’s failure to heed warnings from the AI tool itself.
• Third, the Tribunal underscored the wider significance of its findings on AI misuse, observing that “the issues ventilated in this matter may be of wider assistance”. The Tribunal was clear that it would have imposed the same sanction if either allegation had appeared before it without the other, and independently considered the Rule 14 Allegations on their merits. It found that the Respondent’s misuse of AI breached Principles 1, 2 and 5 through a failure to uphold the rule of law, maintain public trust, and act with integrity, as well as Paragraphs 1.4 and 2.4 of the Code, by misleading the Tribunal and advancing submissions that were not properly arguable. The Tribunal found very high culpability and high harm, and concluded that the AI-related misconduct alone was sufficiently serious to warrant a strike-off order.  The finding of a high level of harm is particularly interesting, as the Tribunal does not appear to have considered it to be a mitigating factor that the misconduct had been drawn to its attention before it had determined the Rule 12 Allegations: “the harm in this matter was the Respondent’s abrogation of his responsibilities as a regulated lawyer.”

Notably, the Tribunal expressly referred to guidance it considered to be publicly available at the time of the misconduct, including the SRA Risk Outlook report on the use of AI in the legal market (November 2023), the SRA's guidance on navigating technology and AI challenges and opportunities (June 2025), and the SRA's compliance tips for solicitors regarding the use of AI and technology (February 2026). The clear implication is that regulated lawyers are now considered to be on notice of the risks associated with AI-generated legal research, and that ignorance of those risks, or of the professional obligations that attend them, is unlikely to afford any defence in future proceedings.