Managing AI generated grievances

Handling an AI-generated grievance is less likely to be overwhelming if managers feel confident about the process they have to follow and about strategies for cutting through the detail to the key issues.

Training is also a core requirement of the draft updated Acas Code.

Training can also help managers spot the signs of an AI-generated grievance and understand the best approach to take.

Requires time and investment at a time when other changes under the Employment Rights Act 2025 are increasing training needs.

This is a long term strategy and won’t address the immediate challenges of a long grievance.

Could require legal input, which would add time and expense to the process.
There are also risks in drawing conclusions before the facts have been investigated.

It’s important that triage is high level and does not delay the process or replace a full and fair investigation.

Adding a meeting at an early stage of the process can be an effective way to cut through the AI noise and get to the heart of the grievance. Given that some complaints may be AI-suggested additions, this will help employers understand which concerns are most strongly held.

Specifically, a scoping meeting could be used to:

• Identify the key concerns;
• Understand which issues are linked or overlapping;
• Explore what the employee sees as a potential resolution; and
• Identify who they would like you to speak to.

This can be summarised in writing (for example in an issues list) and will help focus the process going forward.

This sort of meeting is best held in person. However, many employees are reluctant to attend face to face meetings.

The desired outcome may be unrealistic if AI has given the employee an overly optimistic view of their prospects, so expectations need careful management.

There is no concept of proportionality in the Acas Code, so there is a risk that the employee refers back to their original grievance and argues that it has not been properly dealt with. This risk can be mitigated by offering a right of appeal against the grievance outcome.

An updated grievance policy can reflect how grievances are now raised and anticipate the challenges that AI-generated grievances pose.

Changes could remind employees of responsible AI use, including the risk of sensitive company data, or sensitive personal data about colleagues, being inputted into public AI platforms, and can reinforce or refer to existing acceptable use policies.

A policy could also set out the form a grievance should take ( such as a summary, a chronology and supporting evidence), encouraging employees to focus on facts and proposed solutions rather than legal argument. It could even include a word count limit (see below).

Any requirements on content or structure should operate as guidance rather than a barrier: a grievance that does not follow the format is still a grievance, and the employer's obligations under the Acas Code still apply.

Overly stringent criteria on form and structure could create their own legal risk; for example, the Acas Code makes no provision for restrictions like this. In extreme cases, an employee may claim constructive dismissal for failure to properly consider their grievance, alongside breach of the Acas Code (and the associated compensation uplift).

This can ease the burden on HR of digesting and reviewing extremely lengthy grievances.

Helps to identify the key concerns ahead of a scoping meeting.

Using the business's own enterprise tools, rather than public platforms, reduces confidentiality risk.

Raises data protection considerations in processing the grievance in this way.

A summary may miss or mischaracterise points, so the output must always be checked against the original grievance.

Any summary may well be disclosable in resulting litigation.

Where possible, it is better to keep people at the centre of a difficult grievance process than to fight AI with AI.

Keeps grievances to a more manageable length.

This would be a simple and clear rule to communicate to the workforce. 

Risks being seen as a barrier to raising concerns and lacks nuance for grievances that genuinely require significant detail.

This would not change the employer's obligation to deal with the substance of the grievance, so care is needed to ensure that in practice it is treated as guidance rather than a strict cut off.

This could bring matters to a close quickly, avoiding a lengthy grievance process and potential litigation.

May be particularly helpful where AI has given the employee an overly optimistic view of their prospects, by providing a more realistic forum to discuss outcomes.

Raising settlement too early, or in the wrong way, risks the employee feeling that the grievance is not being taken seriously and that their employer wants to get rid of them.

There are likely to be good arguments that a dispute already exists, so that the discussion is protected by privilege, but this would need to be considered on a case by case basis.

Many employees find AI valuable in helping them organise their thoughts and articulate their concerns. A blanket ban could have a discriminatory impact, as some disabilities make accessing grievance processes more difficult without it.

Sits uneasily with asking employees to adopt generative AI tools at work and given the very widespread use of AI, it would probably be unrealistic and very difficult to police.

May divert attention from the substance of the grievance to how it was written.