The increasing use of artificial intelligence in legal practice has moved rapidly from novelty to routine. In Elden v HMRC [2026] UKFTT 00041 (TC), the First-tier Tribunal (Tax Chamber) was required to confront not only the misuse of AI-generated material in a skeleton argument, but the wider consequences of that misuse for Tribunal resources and professional standards.
While the Tribunal ultimately declined to strike out the appeal, the judgment serves as a pointed warning. Unchecked AI use does not merely prejudice an individual case. It places an unjustifiable burden on the legal process.
The Background
HMRC applied to strike out the appellant’s appeal on the basis that the appellant’s representatives, a firm of regulated accountants, had relied on AI-generated case summaries within their skeleton argument which contained inaccuracies and misstatements of authority.
The application required the Tribunal to consider whether the conduct in question justified the draconian sanction of strike out, and more broadly, how Tribunals should respond to failures arising from the use of AI in legal submissions.
The Legal Framework
Judge Allatt recognised that the use of AI in legal proceedings is a fast-growing area and expressly referenced R (Ayinde) v London Borough of Haringey [2025] EWHC 1383 (Admin). In Ayinde, the High Court confirmed that legal professionals remain under a continuing duty to verify the accuracy of AI-generated legal research before placing it before a court.
Findings of Professional Incompetence
Having reviewed the skeleton argument, Judge Allatt concluded that the case summaries had been produced using AI and had not been checked with sufficient care. The inaccuracies were not minor or technical. They went to the relevance and reliability of the authorities relied upon.
The Tribunal found that this amounted to professional incompetence on the part of any regulated individual or firm involved in the production of the skeleton argument.
Importantly, the criticism was not limited to the initial skeleton argument. The Tribunal was also concerned by SP&C’s conduct after the hearing, particularly their response to the Tribunal’s attempts to clarify why the flawed material had been included.
The Impact on Resources
A central theme of the judgment is the effect of unchecked AI use on public resources. Judge Allatt emphasised that:
- Reviewing irrelevant and inaccurate case summaries occupied a considerable amount of time, which constituted the use of public resources.
- The Tribunal was disappointed by the protracted correspondence following circulation of the draft decision, during which SP&C initially denied any error before later reversing their position.
The judgment makes clear that AI-related errors are not a private inconvenience between parties. They shift the burden onto the Tribunal, undermining its efficiency.
Sanction: Proportionate but Firm
Despite the seriousness of the failings, the Tribunal concluded that strike out would be disproportionate. However, it imposed a series of safeguards aimed at preventing a repeat of the conduct.
In any subsequent skeleton argument:
- Any contributor other than the appellant must confirm which statements of fact or case summaries they have personally checked.
- Those confirmations must be endorsed with a statement of truth.
- Contributors must also provide details of their professional qualifications and regulator.
Judge Allatt made clear that the Tribunal would consider making regulatory referrals where material submitted fell below the professional and ethical standards the Tribunal was entitled to expect.
Practical Implications for Practitioners
This decision reinforces several key principles for those using AI in litigation:
AI is a tool, not an authority. Responsibility for accuracy remains firmly with the professional.
- Inaccurate AI-generated material wastes court time and risks sanctions even where no bad faith is shown.
- Post-hearing conduct matters. A defensive or dismissive attitude towards judicial scrutiny may aggravate the position.
- Tribunals are increasingly willing to impose procedural safeguards and to involve professional regulators where standards fall short.
Is it also a warning for Litigants in Person
Whilst the determination focused heavily on regulated professionals, there is still a lesson here to be learned for Litigants in Person who rely to a greater degree on the use of AI. Whilst AI can assist in understanding the meaning of issues raised, it is by no means accurate and is not a qualified professional. It will make mistakes, and they could be serious and costly.
Conclusion Elden v HMRC is not an anti-AI judgment; rather, it makes clear that any efficiency gains promised by AI cannot come at the expense of accuracy, accountability, or respect for Tribunal resources, and as AI becomes more embedded in legal practice Tribunals are signalling that unchecked use will attract scrutiny not only for what it produces but for what it reveals about individual professional competence, with the message to regulated professionals being unambiguous: where AI assists in preparation, it remains essential that human judgment carries out the checking.
If you wish to discuss any of the above, please contact our Dispute Resolution Team below or call us on 01702 338338.
