ChatGPT drafted it, but you still have to deal with it

Handling AI-generated employee grievances

Employee grievances are nothing new, but how they look has changed. With some initial prompts, what may have been previously raised in a brief email or informal conversation has now transformed into a 50-page essay with references to legislation, legal jargon and HR buzzwords. This may create the illusion of a stronger case. The irony? This often makes it harder for employers to identify the underlying issues that need attention. AI doesn’t verify the facts and often exaggerates or introduces details that were not part of the original complaint.

The figures tell a clear story. A recent survey reported that 70% of UK employers have observed a rise in AI-assisted grievances*. More unsettling, in another survey, 78% of respondents said they felt parts of AI-generated grievances were inaccurate**. 

Spot the difference 

There are some obvious indicators that your employee has used AI instead of writing their grievance on their own: 

• It is repetitive

• It is unnecessarily long 

• It uses American English spelling and Americanisms

• It uses legal jargon and references legislation 

• It uses overly formal language 

• There is an excessive use of dashes

• There is a lack of personal detail 

Why have employees turned to AI? 

The attraction of AI is obvious: it's quick, low-cost, and readily available. It can help employees structure concerns and communicate more confidentially, especially where the employee feels intimidated by the process or doesn’t know where to begin. In many ways, AI is a wonderful tool to help employees have the confidence to raise their concerns.

But, this doesn’t mean every AI-generated grievance has merit. A grievance isn’t stronger simply because it’s well written. Neither is a grievance weaker because it is poorly drafted. 

Responses to the trend 

Australia's Fair Work Commission has recently banned a worker from using AI to prepare further complaints to her employer. Despite having a genuine grievance, the worker’s use of AI intensified the length, scope, and tone used within the correspondence. Australia's Fair Work Commission also noted that the employer's response of effectively disengaging with the AI-generated correspondence meant that some legitimate concerns were not meaningfully addressed. Ignoring correspondence is rarely the safest option. To date, the UK Employment Tribunals have not prohibited the use of AI-assistance. However, amongst the growing trend in AI grievances and the pitfalls that employers face in dealing with them, it would likely be a move that is well-received by employers. 

In the UK, ACAS has received feedback about an increasing tendency for written grievances to be overly lengthy and complex. This trend can cause delays and, more often than not, heighten tensions between the parties. To address this, ACAS has recently published a revised draft Code of Practice on Disciplinary and Grievance Procedures: Code of Practice on disciplinary and grievance procedures (draft) | Acas, which is currently open for consultation.

The draft Code encourages employees to set out their concerns in a concise and straightforward written statement at the outset. It also sets out that employers should treat such concerns seriously and provide employees with the opportunity to expand on or clarify their concerns, where necessary.  

How can employers tackle the burden? 

Whilst it’s tempting to make an outright ban on AI-generated grievances, the ACAS Code of Practice requires employers to investigate all grievances, no matter how they were prepared. A fair process must still be followed as with any other grievance, but this can be costly and time-consuming for employers when there is a lot of filler text. 

This doesn’t mean that employers need to reply to each line of the employee’s grievance in turn. Instead, employers should be focusing their efforts to establish facts and investigate substance.

A helpful starting point is to:

• extract background details 

• determine allegations that need further investigation

• match the issues to the relevant internal policies

From there, employers should determine whether: 

• the issues have been addressed previously 

• there are any witnesses or evidence to the issues raised 

• there are any apparent cost-effective solutions to the issues raised 

Best practice for employers

In response to AI-generated grievances, employers should make sure to:

• Identify and clarify the employee's main concerns. Face-to-face meetings can help to recognise the matters that go to the heart of the employee's complaint and keep communication open.

• Provide training to managers on how to identify grievances and respond appropriately. This helps to avoid the business dismissing legitimate concerns and ensures matters are addressed fairly and consistently.

• Keep workplace policies up to date. Employees should have clear pathways to raise concerns. Informal conversations are always a good starting point to help mitigate the risk of both AI-generated and handwritten grievances in the first place.

• Provide training to employees about the risks and consequences of inserting company-sensitive information into open-source AI. This may act as a deterrent to using AI to prepare grievances in itself. 

Grievance handling is still ultimately about people. Employees expect to be heard, understood and treated fairly. No matter how sophisticated the technology becomes, those expectations will remain.

If you are facing an employee grievance and need some support with resolving it, speak to our Employment Law Team today.

 *WorkNest, “Seven in ten employers report rise in AI-assisted grievances as manager confidence remains low” (WorkNest, July 2026) <https://worknest.com/blog/rise-in-ai-assisted-grievances-as-manager-confidence-remains-low>
 

**Mahalia Mayne, “AI-assisted grievances pile pressure on people teams, survey finds” (People Management, 25 February 2026) <https://www.peoplemanagement.co.uk/article/1949582/ai-assisted-grievances-pile-pressure-people-teams-survey-finds>