Generative AI can make a lawyer’s life easier; it can also end a legal career in the space of one regulatory filing. Abhishek Kumar, a 41-year-old registered foreign lawyer, discovered exactly how high the price of carelessness can be. After using generative AI to help prepare his written response in disciplinary proceedings brought by the Solicitors Regulation Authority, Kumar filed material that contained false and misleading case citations. When the regulator flagged the errors, he used AI again to explain what had happened — and produced yet more errors. The Solicitors Disciplinary Tribunal has now struck him off.
The decision is not simply a story about a lawyer with a criminal record facing professional consequences. It is one of the clearest signals yet that a regulated lawyer cannot blame generative AI for unreliable legal content. The tribunal found Kumar responsible for the accuracy of the material placed before both the SRA and the tribunal, describing his culpability for the AI-related misconduct as “very high”.
What makes the case especially significant is the tribunal’s conclusion that Kumar would have been struck off on the basis of either the criminal conviction or the AI-related misconduct alone. In other words, even setting aside the underlying criminal matter, the way he used generative AI in his own defence was sufficient to justify removing him from the register of foreign lawyers.
The Criminal Conviction That Brought a Registered Foreign Lawyer Before the Tribunal
The disciplinary case began with an event that had nothing to do with artificial intelligence. Kumar had previously represented an immigration client. That client later carried out administrative work for him, was paid cash in hand, and had no legal right to work in the UK. In 2024, Kumar was convicted in connection with knowingly employing that person and received a 12-month community order together with 150 hours of unpaid work.
For a regulated individual, a criminal conviction of this kind is automatically a serious professional matter. It calls into question a lawyer’s willingness to comply with the law and to act with integrity. The SRA, which regulates solicitors and law firms in England and Wales, brought disciplinary proceedings against Kumar and referred him to the Solicitors Disciplinary Tribunal.
Kumar had been a registered foreign lawyer since 2018. That status allows a lawyer qualified outside England and Wales to practise in English and Welsh law through an SRA-regulated firm. But registration does not mean a lighter set of obligations. On the contrary, registered foreign lawyers are expected to comply with the SRA’s professional standards, including the duty to act with honesty, integrity, and competence in everything they file.
A Lawyer Uses Fake AI Cases at a Disciplinary Tribunal — and Is Struck Off
The AI-related misconduct arose after the SRA’s proceedings were already underway. Kumar was required to set out his response in writing, and he turned to generative AI to help draft it. The response he submitted included case citations that were not reliable. The tribunal found that they were false and misleading in multiple ways: some citations were wrong or pointed to unrelated cases, while other authorities did not support the legal points he was trying to make.
The SRA identified the problems and asked Kumar to explain. Instead of going back to the original sources and checking the citations by hand, Kumar again relied on generative AI to prepare his explanation. That explanation, the tribunal found, contained further errors. By that point, the issue was no longer an isolated mistake; it had become a repeated failure of professional judgment.
Kumar admitted that he had used generative AI to assist with drafting the initial document. But the admission did not reduce the seriousness of what had happened. The tribunal found that Kumar had failed to properly check the material he was putting before both the regulator and the tribunal, and that his culpability for the AI-related misconduct was “very high”.
Why did the tribunal reject Kumar’s claim that the mistakes were just incompetence?
The tribunal did not accept that inexperience with artificial intelligence excused the failure. As a regulated lawyer, Kumar was obliged to comply with professional standards, and those standards do not change when drafting is moved to a chatbot. He remained responsible for every case citation in his own defence, whether or not he understood exactly how the technology generated it. Because he used AI twice without properly verifying the output, the tribunal concluded that the repeated failure carried serious implications for the administration of justice and public confidence in the legal profession.
Kumar told the tribunal that he had not intended to mislead anyone. He argued that the errors reflected incompetence and poor judgment rather than dishonesty or a lack of integrity. He also said that he was accustomed to working under supervision and did not have the expertise needed to properly verify AI-generated output. The tribunal took a different view. It said that a regulated lawyer could not outsource professional responsibilities to an AI system, and that the duty to ensure accuracy remained with the lawyer regardless of how the text was produced.
In many ways, the tribunal’s response to Kumar’s supervision argument is the most important element of the case. Kumar suggested that because he normally worked under someone else’s supervision, he lacked the skills to check the AI’s output. But the tribunal made clear that a lawyer’s professional obligations are not contingent on how much supervision they receive. The person supervising the AI in this case was Kumar himself, and he failed to exercise that supervisory responsibility at all.
The Second Use of AI That Made the Case Worse
The tribunal placed significant weight on the fact that Kumar used AI a second time when responding to the mistakes created by his first use of the technology. That second use was not merely an attempt to clarify; it produced more false and misleading material and effectively compounded the original error.
This detail matters because it distinguishes a genuine accident from a pattern of conduct. A lawyer who uses AI once and then discovers that the citations are invented can argue that the first mistake was a one-off failure. A lawyer who uses AI again to explain the first failure, without any check on the new output, demonstrates a continuing disregard for the reliability of what he is filing. In Kumar’s case, the tribunal concluded that this was not just carelessness; it was a fundamental failure to understand the duties of a regulated lawyer.
The tribunal also said that Kumar had used AI “without giving any proper thought to the consequences”. That is a striking phrase. It suggests that the problem was not the technology itself, but the complete absence of judgment about how the technology might fail. Any lawyer who uses generative AI knows that it can produce confident-sounding but entirely incorrect legal authorities. The duty to check what a machine produces is not a technical skill; it is the same core legal skill of verifying that one’s sources actually support one’s arguments.
What This Striking-Off Means for Lawyers Who Use Generative AI
For the legal profession, the case is an emphatic warning: the fact that a document was created by artificial intelligence does not make a lawyer any less responsible for it. The SDT’s decision confirms that regulators will treat AI-generated false citations as a serious conduct issue, not as a technological glitch to be forgiven.
The most immediate lesson is practical. Every word in a filing, and above all every case citation, must be independently verified against an original source. Generative AI sometimes produces authorities that do not exist, misstates the holdings of real cases, or attributes legal principles to the wrong court. The only reliable safeguard is to look up the case, read it, and confirm that it supports the proposition for which it is cited. If a citation cannot be verified, it should not be used.
The second lesson is disciplinary. Lawyers are personally responsible for what they file, even when the text was generated by software. That principle has long been at the heart of professional regulation in England and Wales, and Kumar’s case confirms that it applies with full force to AI-assisted legal work. A lawyer cannot argue that the AI made a mistake, because the AI is not the lawyer; the lawyer is the one who signed and submitted the document.
The third lesson is institutional. Law firms that permit or encourage the use of generative AI need clear policies, training, and review workflows. When a lawyer uses AI to draft submissions and fails to verify the output, the immediate risk is personal sanctions. But the wider risk extends to firms, clients, and the administration of justice. A firm that allows AI-assisted drafting without clear rules on verification is exposing itself to regulatory damage that can be difficult to repair.
The case also raises a deeper question about professional identity. Lawyering has always involved the exercise of judgment: deciding which authorities matter, whether an argument is supported by the record, and whether a claim can be made in good faith. Generative AI can assist with some of this work, but it cannot replace the judgment required to take responsibility for the finished product. When a lawyer surrenders that judgment to a machine, they are not merely making a technological error; they are abandoning the core of what it means to practise law.
Kumar’s removal from the register of foreign lawyers is therefore more than a personal punishment. It is a statement about the standards that must continue to govern the profession as legal technology advances. The tribunal’s decision makes clear that the old rules still apply in the new AI era: lawyers may use whatever tools they like, but they remain answerable for every case, every citation, and every claim placed before a court or regulator.
The confidence of the public, the courts, and the profession itself depends on that principle being enforced. Kumar’s case shows that, in the eyes of the Solicitors Disciplinary Tribunal, no amount of technological convenience can replace the oldest obligation in law: making sure that what is filed is true, accurate, and fit to be relied upon.
