EAT Guidance on the use of AI by Litigants-in-Person: Hancox v Sutherland & Others [2026] EAT 139

Employment practitioners will by now be familiar with responding to AI generated claims brought by litigants-in-person (LiPs). There is no denying the potential of AI to expand access to justice for LiPs. However, the reality of their AI use has been a marked increase in the volume of the material they produce in the course of proceedings.

In September, the EAT handed down its decision in Hancox v Sutherland & Others [2026] EAT 139. In its judgment, particularly at paragraphs 6–37, the EAT gives useful guidance on the use of AI by LiPs in Employment litigation. This article sets out that guidance as well as some other useful sources for managing AI use by LiPs.

In Hancox, two days before the date of a preliminary hearing in the EAT, the appellant, Mr Hancox (‘H’), filed a 300 page skeleton argument (totalling nearly 132,000 words) which had been generated by ChatGPT. In the covering email to which his skeleton argument was attached, H stated:

“The Respondents now have the consolidated document and therefore have a reasonable opportunity before Wednesday’s hearing to identify any specific:

• factual inaccuracy;
• incorrect quotation;
• mistaken date or time;
• inaccurate description of a document;
• erroneous bundle reference; or
• material passage which they say misrepresents their position.

I will consider any properly particularised correction promptly and, where appropriate, provide a correction or clarification before the hearing.” [83]

The EAT described H’s skeleton argument as entirely unacceptable. Not only did it exceed the page limit prescribed by the Practice Direction to the EAT (‘PD’), but it also referred to documents which were not in the bundle [84]. The EAT also strongly objected to H placing the onus on R to identify inaccuracies [86].

Guidance on LiP AI use from Hancox

In arriving at its assessment of H’s conduct, the EAT provided some useful general guidance on the use of AI by LiPs.

At paragraphs 6 to 21, specific to the EAT:

• LiPs are expected to familiarise themselves with the EAT Rules and the PD. These are comprehensible to people without legal qualifications and are available to access on the internet [8].
• The stress of bringing an appeal is not generally a good reason for failing to cooperate with the EAT and other parties [13].
• Compliance with the skeleton argument rules set out in the PD is mandatory for all litigants. Being a LiP or being unaware of the PD is not generally a good reason for non-compliance. It logically follows that the use of AI is not a good reason for non-compliance [19].
• All litigants, including LiPs, should expect the EAT to use its powers to enforce compliance with the EAT Rules and the PD. Skeleton arguments which are too long may be rejected and a shorter version ordered. The EAT can also reduce time for oral submission on the basis that the written arguments are extensive [21].

Then, at paragraphs 22-37, in relation to employment litigation generally:

• AI use in litigation risks material being put before courts and tribunals that is unfocused, irrelevant and legally or factually incorrect [23].
• There is no reason why LiPs cannot be expected to take reasonable steps to ensure they are using AI responsibly. The underlying principles from R (Ayinde) v London Borough of Haringey [2025] EWHC 1383 (Admin) apply to all persons who submit documents to a court or tribunal, whether legally qualified or not [31].
• LiPs who use AI should ensure, as a minimum, that documents they submit:
  • Comply with applicable procedural rules;
  • Have been checked as thoroughly as the LiP is reasonably able for accuracy, ensuring that factual, evidential and legal points (including references to authorities) are correct and that the court or tribunal is not misled; and
  • Contain only relevant points, with the focus being on the central or best arguments, presented in an easily comprehensible manner and avoiding undue repetition [32].
• Submitting AI generated material to a court or tribunal and simply expecting opponents to check it for accuracy is not an acceptable approach. Where it is apparent that the necessary checks for accuracy have not been undertaken, or have been done inadequately, sanctions are likely to ensue [33].
• Judicial responses to inappropriate AI use by LiPs can include, in the most serious cases, a referral to the police to investigate the common law offence of perverting the course of justice, or the initiation of contempt proceedings. In less serious cases the normal range of sanctions (e.g. strike out, costs and deposit orders) is applicable [35].
• The use of AI in preparing witness statements can result in a cautious approach to the evidence (as it did in Godwin v Godwin [2026] EWHC 923 (Ch)). Use of AI whilst giving evidence can amount to witness coaching and thus is not permitted. Where factual errors in evidence are blamed on AI error this will require careful scrutiny [36].
• All litigants must take personal responsibility for any document submitted to the Tribunal and it must be checked in the ways described in the above guidance, as thoroughly as the author is reasonably able [37].

Paragraphs 22–37 of Hancox are likely to be useful to parties seeking to control inappropriate AI use in the ET. A party should identify the putative disproportionate AI use and remind the LiP of the responsibilities as set out in Hancox. If the problem is not addressed, the party can then draw those paragraphs of Hancox to the Tribunal’s attention and seek directions to address the LiP’s conduct.

LiPs should be expected to be aware of the ET Procedure Rules 2024, including their obligation to act in accordance with the Overriding Objective in relation to their AI use. Parties seeking to acquaint LiPs with these obligations may find a reference to paragraph 32 of Hancox (see above) particularly useful here, especially when accompanied by a reminder that they may face sanctions if they do not regulate their use of AI.

Other potentially useful sources for managing AI use by LiPs

In October 2025 the ‘Artificial Intelligence – Guidance for Judicial Office Holders’ guidance was released. In it reference is made to the use of AI by unrepresented litigants:

“Litigants rarely have the skills independently to verify legal information provided by AI chatbots and may not be aware that they are prone to error. If it appears an AI chatbot may have been used to prepare submissions or other documents, it is appropriate to inquire about this, ask what checks for accuracy have been undertaken (if any), and inform the litigant that they are responsible for what they put to the court/tribunal.” (emphasis added)

Further, in June 2026 Presidential Guidance on applications for interim relief was released, following a sharp increase in the number of AI-generated applications for interim relief. At paragraph 29 the Guidance says the following: AI “often results in submissions that are too long and complex, contain irrelevant material and fail to focus on the key points in the case. Litigants who use AI to assist them have a responsibility to ensure that what is submitted is concise, relevant and accurate.” (emphasis added)

These sources are useful when read in conjunction with paragraphs 32 and 33 of Hancox (see above). LiPs should be made aware that their responsibility for the material they submit extends to checking thoroughly for accuracy, focusing on their best points and avoiding repetition. Tribunals should be reminded that a LiP has these obligations if and when the issue of AI use is raised.

Some practitioners will already be aware that Bristol ET has released a direction on the management of AI use by parties. It states that if a party generates material using AI in the course of litigation it must:

• State at the head of the document that AI has been used;
• Confirm that it has been checked for accuracy (both as to content and reference to any legal provisions or case law);
• Confirm that it has been checked for relevance;
• Confirm that the document has been reviewed to ensure it is proportionate in length and that the content is not repetitive;
• Confirm that the overriding objective has been considered before any correspondence, application or document is submitted to the Tribunal or sent to the other party.

The direction also states that if a Judge decides that:

• Any argument in such a document is entirely without merit so that pursuing the argument is unreasonable, and/or
• The party does not understand the argument and so cannot explain it to the Judge with the result that the party’s conduct in pursuing the argument is unreasonable, and/or
• The document contains references to a legal case which does not exist or suggests that a legal case establishes a legal principle which it does not,

the judge is entitled to conclude that the party’s conduct is unreasonable and may impose sanctions, including strike out and/or deposit orders. Bristol’s more radical approach has not yet been replicated in other regions. The extent to which it can be used to persuade Tribunals elsewhere to impose sanctions is unclear.

Finally, the Civil Justice Council released an interim report entitled ‘Use of AI for Preparing Court Documents’ in February 2026. The Report is focused on the use of AI by legal professionals, and only really pays lip service to LiPs at paragraphs 2.21-2.23. The Report does recognise the risks involved with LiP AI use, and suggests it might be useful for AI generated documents to be labelled as such, but expresses no concluded view. It is hoped that the Final Report, planned for release later this year, will give further guidance when it is released.

Comment

Hancox and the above sources of guidance are useful tools with which to reduce the time spent and costs incurred litigating against LiPs who use AI in the course of proceedings. They should help to ensure that LiPs check the accuracy of the material they produce and at least try to ensure the volume they file is proportionate and relevant.

However, a coherent approach in the form of Presidential Guidance which regulates AI use by LiPs would clearly help further. Respondent practitioners will be aware that the sanctions warnings they issue to Claimants via correspondence can often fall on deaf ears. The lack of clarity inherent to the piecemeal approach taken so far to the regulation of AI use by LiPs arguably exacerbates this. Presidential Guidance would make it easier for parties to remind their in-person opponents of their responsibilities when it comes to AI. Neutral guidance from the Tribunal is more likely to be trusted and followed by LiPs. It would be simpler for Tribunals to apply and may encourage judges to be more willing to manage AI use from the outset of proceedings. Tribunals may also be more willing to apply a Hancox style approach with ET specific guidance. It may also be the case that Guidance helps the AI technology itself to produce material more appropriate to Tribunal proceedings.

Ultimately, no amount of guidance is going to solve all of the problems created by LiP use of AI. AI generated documents are almost inherently going to be longer and thus more costly to respond to. Even with Hancox and any future guidance, LiPs are unlikely to face sanctions for misidentifying legal principles from authorities. It is also unlikely that LiPs will be able to identify whether AI generated material is relevant in all cases, and thus it is hard to envisage sanctions following in every case. Questions remain over whether it would be in the interests of justice to sanction a LiP for relying on AI generated material which they do not understand but nonetheless has merit.

Hancox is nonetheless a good starting point. Given the impact that AI is having on Employment litigation, I think we are very likely to see further guidance on its use by LiPs. We will have to wait and see what form this will take. In the meantime, the passages from Hancox and other sources I have listed above should at least begin to help parties manage AI use by LiPs.

Samuel is a first six pupil in the Employment team under the supervision of Louise Quigley. He will be available for instructions from March 2027.

Sources

Hancox v Sutherland & Others [2026] EAT 139: https://www.gov.uk/employment-appeal-tribunal-decisions/mr-t-hancox-v-kenneth-sutherland-and-others-2026-eat-139

Presidential Guidance – Applications for interim relief – June 2026: https://www.judiciary.uk/wp-content/uploads/2026/06/PG-on-IRF-June-2026-final.pdf

Artificial Intelligence, Guidance for Judicial Office Holders – October 2025: https://www.judiciary.uk/guidance-and-resources/artificial-intelligence-ai-judicial-guidance-october-2025/

Use of AI for Preparing Court Documents – Interim Report and Consultation. Civil Justice Council – February 2026: https://www.judiciary.uk/wp-content/uploads/2026/02/Interim-Report-and-Consultation-Use-of-AI-for-Preparing-Court-Documents-2.pdf