Employment Appeal Tribunal ruling on AI-drafted court documents: what litigants in person face from September 2026

A deputy High Court judge sitting in the Employment Appeal Tribunal (EAT) issued a pointed warning in September 2026 that should concentrate minds well beyond the courtroom. Marcus Pilgerstorfer KC told litigants in person that submitting AI-generated court documents without thorough personal checks is unacceptable — and that doing so risks judicial inquiry and sanctions.

The ruling did not ban AI from the drafting process. What it made clear is that responsibility for the content of any document submitted to a court rests entirely with the person who submits it. That distinction matters enormously for professional services businesses advising clients, managing employment disputes, or operating in jurisdictions where courts are actively tightening their expectations around AI use.

What the EAT Actually Said

Pilgerstorfer KC's remarks were directed at litigants in person — individuals who represent themselves without a solicitor. But the principle underlying his warning applies far more broadly. Courts are not interested in how a document was drafted. They are interested in its accuracy, its truthfulness, and whether it fairly represents the case being argued.

AI tools can generate plausible-sounding legal text that contains fabricated case citations, misrepresents legal standards, or simply gets the facts wrong. A litigant who submits such a document without checking it has, in the eyes of the court, submitted that content personally. Ignorance of how an AI tool works is not a defence.

For professional services firms, the practical read-through is this: if your staff are using AI to draft correspondence, instructions to counsel, witness statements, or any document that may eventually find its way into legal proceedings, you need a verification process that can withstand scrutiny.

Why This Matters Beyond the UK

UK case law and regulatory guidance increasingly inform how other common law jurisdictions approach similar questions. Australia, Canada, Singapore, and several US federal courts have already issued or are developing their own guidance on AI-generated submissions. The EAT's position aligns with a growing international consensus: courts will hold submitting parties to the same standard regardless of whether a human or an AI produced the first draft.

For firms operating across multiple jurisdictions — international accounting networks, global HR consultancies, law firms with cross-border practices — this is not a UK-specific compliance issue. It is a global one. Policies that govern how AI is used in document production need to be consistent and auditable, not left to individual judgement at the point of submission.

Marketing agencies drafting client-facing regulatory documents, HR consultancies producing employment tribunal submissions on behalf of clients, and accountants generating correspondence with HMRC or overseas tax authorities are all in scope. If the document could end up before a regulator, a court, or a tribunal, your AI governance policy needs to speak to it.

The Broader UK Regulatory Picture

The EAT warning did not arrive in isolation. September 2026 has brought a cluster of significant developments that together signal where UK AI governance is heading.

Parliament's Joint Committee on Human Rights recommended in mid-September that the government introduce a risk-based AI Bill and establish an independent statutory regulator. This is a committee recommendation, not yet legislation or government policy, but committee recommendations of this kind tend to shape what follows. Firms that wait for legislation before building compliance frameworks typically find themselves scrambling.

At the UN General Assembly, Prime Minister Andy Burnham announced that the UK would use its 2027 G20 presidency to champion global principles and standards for frontier AI development. UK officials have signalled to AI companies that full visibility over new models is expected, with potential legislation to mandate transparency. The House of Commons Business, Innovation, Science and Trade Committee has called AI firms to an urgent hearing in October.

The direction is unambiguous. Greater oversight, greater transparency requirements, and greater accountability for AI outputs are coming. The question is not whether your business will need to demonstrate compliance — it is whether you will be ready when the obligation crystallises.

What Professional Services Firms Should Do Now

The EAT's warning is a useful prompt for a broader compliance audit. Here is where to focus.

Document production workflows. Map every stage at which AI tools are used to produce documents that may carry legal, regulatory, or reputational weight. Identify who has responsibility for final sign-off and whether that person has the knowledge and time to conduct a genuine check — not a cursory read.

Verification standards. Establish a written standard for what constitutes adequate review of AI-generated content. This should specify what categories of content require independent verification (legal citations, factual claims, regulatory references), by whom, and how that verification is recorded.

Staff training. Ensure that everyone using AI drafting tools understands that they are personally accountable for the content they submit or approve. Training should cover the specific failure modes of the tools in use, not just general AI literacy.

Policy documentation. If your firm operates across multiple jurisdictions, your AI use policy needs to reflect the requirements of each. A single-page acceptable use policy is unlikely to be sufficient. Courts and regulators in different jurisdictions are developing their own expectations, and your documentation needs to demonstrate that you have engaged with those requirements seriously.

Supplier due diligence. If you are using third-party AI tools to support client work, you need to understand the terms under which those tools operate, what data they process, and what representations the supplier makes about accuracy. The EAT ruling makes clear that outsourcing the drafting to an AI tool does not outsource the responsibility.

The Compliance Gap Is Closing

The gap between what businesses are doing with AI and what regulators and courts expect of them is narrowing, quickly. The EAT's September 2026 warning is not an outlier — it reflects a sustained effort by courts across common law jurisdictions to set clear expectations before AI-generated content becomes so embedded in legal proceedings that the problems become structural.

For professional services firms, the reputational and financial consequences of getting this wrong are significant. A sanctions finding, a wasted costs order, or a regulatory referral arising from inadequate AI governance is not an abstract risk. It is the kind of event that affects client relationships and insurance premiums in ways that take years to recover from.

The businesses that will navigate this period well are those that treat AI governance as an operational priority now, rather than a compliance exercise to be addressed after something goes wrong.

How Ops Intel Can Help

Ops Intel works with professional services firms globally to build AI compliance frameworks that are proportionate, auditable, and designed to hold up under scrutiny — from courts, regulators, and clients alike.

If the developments outlined above have raised questions about your current AI governance arrangements, we would welcome a conversation.

Visit https://www.opsintel.io to learn more about our compliance services or to get in touch with our team.