Fighting back – tackling the use of generative AI in the courts and the workplace

Sally Bird, Partner

In this article I take a look at what the steps that the courts and employers may take to mitigate some of the evils of AI in employment related matters.

There is no doubt that 2026 has seen a huge increase in the use and impact of AI both within workplaces and the Courts and tribunals. As its development and use continue to grow exponentially this trend is only set to continue. Through its tendency to elongate, over-complicate, misrepresent or outright hallucinate AI is adding significantly to the workload of tribunals, lawyers and employers and not in a good way.

It is the favoured tool of the vexatious litigant in person or disgruntled employee which uses up valuable court time and resources, management and HR time and escalates legal costs. Coupled with the widespread use of Data Subject Access Requests it is leading to vastly lengthy and complex internal grievances, which can feel like full scale litigation even before a claim has been issued. What then can courts and employers do to tackle this?

Court guidance

This week the Employment Appeal Tribunal (EAT) has given guidance on the use of generative AI in preparation of documents.  This arose from a case involving an employee without legal representation who was appealing against the strike out of his claim by the Employment Tribunal. He appealed against this decision and filed a 300-page skeleton argument which he had created using ChatGPT. Unbelievably, the employee then asked his employer to go through this to identify any factual errors, inaccurate quotations, incorrect dates, faulty bundle references or misrepresentations.

The EAT found the document “entirely unacceptable”, and its statement that AI served to “obscure rather than to illuminate” the issues really captures the whole point. The practice direction’s recommended page limit was 5-15 pages, not 300 and it was plainly not acceptable that the employee had not checked the document. The EAT noted the significant increase in AI usage before courts and tribunals, and that while responsible and ethical use of AI could benefit unrepresented parties, its output may be unfocused, excessive or misleading or contain hallucinations, and noted that readily available tools such as ChatGPT could not conduct reliable research. It underlined that litigants in person should use AI responsibly and that the principles of personal responsibility and accuracy applied equally to litigants in person as well as to legal professionals.

In particular it made clear that litigants who use AI must ensure, as a minimum, that all documents submitted:

• Comply with applicable procedural rules (including page or word limits)
• Have been checked as thoroughly as reasonably possible for accuracy, ensuring that factual, evidential and legal points are correct and that courts and tribunals are not mislead
• Contain only relevant points, focussed on central or best arguments and are presented in an easily comprehensible manner without undue repetition

The submission of documents without checking them or placing the onus to check on another party if not acceptable – and where this is apparent the EAT warned that potential sanctions could follow.

This guidance from the EAT is welcome and will be particularly helpful in empowering employers and their lawyers to challenge the misuse of AI in tribunal proceedings.

Updates to grievance procedures

The EAT guidance begs the question how then might employers import this useful guidance into their own internal procedures to introduce some checks and balances on the use of AI by employees?

We have all seen that the increased use of AI to draft grievance letters results in longer grievances that lack focus and coherence, are repetitive and make it difficult to actually understand the core of the employee’s actual complaint and what resolution they are looking for.

Another serious issue which arises from the increased use of AI is that often employees do not realise that AI tools are not private and confidential, and the uploading of confidential, sensitive information or personal data of others exposes the organisation to risks and places them in breach of confidentiality and data protection obligations.

One good way of tackling these issues is to update Grievance procedures so that they impose clear limitations and include measures designed to focus the issues and help employees to articulate their grievances in simple and clear terms. This could include:

• Introducing parameters on the length of grievances -imposing a maximum word limit or page limit
• Making clear that employees should not enter confidential information or information which discloses the personal data of anyone other than them into an AI tool, which would be treated as a serious breach of confidentiality for which disciplinary action could be taken
• Requiring or encouraging employees to draft grievances in their own words and making clear that if they do contain any AI generated content it should be checked carefully for accuracy
• Reserving the right to hold a scoping meeting requiring the employee to put their grievance in their own words. identifying the key issues that are raised, leading to a grievance summary being produced so that focus can be kept on the key issues
• Considering the introduction of a template form on which grievances could be raised including a preferred structure

How we can assist

We can draft a bespoke grievance procedure incorporating some or all of the suggestions mentioned above. We can also provide guidance and offer training for managers in handling AI generated grievances, including the safeguards that should be followed in their own use of AI when investigating and assessing grievances. Please get in touch with our employment team at [email protected] or 0161 832 3434.