For January 2026, we have asked the employment team at Tughans LLP to provide practical answers to unusual, sensitive or complex work-related queries. We call this feature “How do I handle it?”

The articles are aimed at HR professionals and other managers who may need to deal, from time to time, with the less commonplace disputes at work; issues that may, if handled incorrectly, lead to claims for discrimination, constructive dismissal or some other serious difficulty. 
 

As use of generative AI platforms like ChatGPT continues to grow, employees are increasing lodging grievances which have been largely or entirely AI generated. They are often extremely detailed and complex, even when related to relatively simple or discrete issues. The longest I have seen so far was 60 pages – complete with its own contents page!

Depending on the prompts used by the employee, AI grievances can be hard to follow, without clearly identifying the employee’s complaints, their evidence or the outcome they want. They can use language which the employee would not ordinarily use themselves, and which they might not fully intend. They commonly include threats of further action through early conciliation, legal claims, subject access requests or demands for compensation. AI grievances can naturally become a significant burden and are difficult to deal with effectively while maintaining the employment relationship.
 
Despite this, an AI grievance is a grievance nonetheless and cannot be disregarded. You will need to complete a reasonable and fair investigation in the normal way, in accordance with your internal grievance procedure and the LRA Code of Practice. 

The length and complexity of the grievance will make the initial grievance hearing more important. This is an opportunity to go through the grievance with the employee. You will be able to ask them to explain any areas which are unclear and ask them what they mean by certain words or phrases used. This will allow you to address any language which seems out of place, overly combative or emotive. In some cases, employees may have submitted the AI grievance without fully considering the language they are using and if it accurately reflects what they want to say.
 
The grievance hearing will also be the main opportunity to explore other problems with the grievance contents. While AI grievances can appear to be sophisticated and well developed at first glance, they will often include inaccuracies or irrelevant information. For example, they may link the facts of the employee’s case to the wrong type of complaint or legislation or request a solution which is not directly relevant or that you cannot provide. They can refer to cases or laws from different countries, or that have been “hallucinated” and do not actually exist. This is a particular problem in Northern Ireland, given the relatively small amount of information about NI employment law and practice for AI tools to “train” on, and grievances will often refer to a mixture of GB and other laws.

In our experience, AI grievances often include extensive and grossly exaggerated requests for compensation, often based on increasingly bizarre forms of losses which cannot be claimed in the courts or tribunals. 

This is problematic firstly because compensation is not a valid outcome to an internal grievance. Secondly, AI grievances can greatly distort the employee’s view of their complaints and expectations on what a fair outcome should look like. 

Ideally, the employee will cooperate and explain their grievance in their own words. However, they may refuse and indicate that they have included everything in the AI grievance submission and have nothing further to add. You should take a careful note of the meeting, as their attitude may be relevant factor in any later tribunal proceedings. It is likely (and rather hoped) that an employee’s continued unreasonable or vexatious use of AI during an internal process will soon be scrutinized by a court or tribunal.

You will then need to work through the grievance. This will involve trying to isolate and identify the relevant complaints, and any evidence that has been provided, so that you can investigate and reach outcomes on them. You should also identify any complaints or demands which cannot be addressed within the grievance process, so that this can be noted in the grievance outcome. You should watch for any requests for documentation or personal information which should be properly treated as a data subject access request.

If you have not already, you may wish to implement a policy on using generative AI in the workplace, as well as training staff on how AI works, its benefits and limitations. While this will not prevent employees from submitting AI grievances, it may deter them from placing “blind faith” in what it produces and encourage responsible use.
 
The most enduring impact of AI is likely to be the erosion of trust between employee and management during grievances and other contentious internal processes. Employees may regard AI as a neutral and unimpeachable source of information about their rights and entitlements and consequently will trust the AI over advice provided to them by HR or their managers. This can be very unhelpful, eroding trust and frustrating your ability to resolve issues in a proportionate, amicable way. 

Overall, you should take the steps which are reasonably available to you to tackle the grievance, and where aspects are unclear, seek further input from the employee, before giving your response. If the employee is unable to explain what a particular part of their grievances means, or was intended to mean, or has made demands which you cannot fulfil, this will naturally inform the outcome you reach.