AI has changed the dynamic of workplace grievances. And not for the better. One imprecise prompt can now turn a relatively simple complaint into a lengthy, legalistic grievance in minutes. Some contain genuine issues expressed more clearly. Others arrive padded with irrelevant law and allegations the employee may struggle to explain. And the repetition. The constant repetition.

The danger is not the technology. It is allowing the daunting number of words to dictate how your organisation responds.

What is changing?

AI-assisted grievances are now sufficiently prominent that Acas has expressly raised AI use in its current consultation on a revised disciplinary and grievance Code of Practice. Its draft also encourages workers to keep written grievances short and clear and, where possible, say what resolution they want.

That matters. The emerging problem is not simply more complaints. The challenge is that AI makes escalation easy. Uncertainty and frustration can turn a manageable disagreement into something that looks like litigation before anyone has properly discussed what happened.

What does this mean in practice?

A 30-page grievance creates an obvious temptation. Feed an AI-generated grievance into another AI tool and let the bots battle it out.

An employee using AI to help draft a grievance is not a problem. It can help those struggling to organise and articulate their thoughts.  Particularly if English is not their first language.

But the substance of the complaint matters, not who or what helped with the wording. Consider how often your lawyers draft your correspondence.  The important questions are: What are they alleging? What evidence supports it? What do they want you to do about it?

That becomes a governance issue when allegations concern discrimination, harassment, whistleblowing, senior misconduct or retaliation. You need to distinguish material allegations from drafting clutter, escalate serious issues and preserve reliable evidence.

If the dispute reaches a Tribunal, polished prose will matter much less than contemporaneous evidence. Can you show who made the decisions, what they knew, what they investigated and why they considered their response reasonable? A beautifully formatted grievance does not cure weak facts. Neither does a beautifully formatted employer response.

So, what should you do?

• Triage substance, not style. Identify the factual allegations, people involved, relevant dates and requested outcome. Don’t spend hours debating whether ChatGPT wrote the document.
• Ask the employee to clarify. Where allegations are vague, repetitive or contradictory, ask focused questions. A grievance meeting can quickly reveal which issues the employee genuinely advances and which have simply found their way into the document.
• Define the investigation scope. Record what you will investigate and why. Do not assume every paragraph requires a separate finding. But do not quietly discard an allegation because it looks exaggerated.
• Escalate the serious issues. Set clear triggers for allegations involving whistleblowing, discrimination, harassment, fraud, regulatory breaches or senior leaders. These can create risk well beyond an ordinary employee relations dispute.
• Protect the evidence. Preserve emails, messages, meeting notes and decision records early. Memory becomes remarkably creative once positions harden.
• Control your own AI use. Do not put sensitive grievance material into unapproved public tools. Check confidentiality, privilege and data protection before using AI to summarise complaints or draft outcomes.
• Train managers to resolve problems earlier. The cheapest grievance to investigate is the one that never becomes formal. Give managers the confidence and authority to address concerns before positions become entrenched.

AI makes it easier to produce formal complaints. Don’t respond with more paperwork.  Respond with sharper judgement, disciplined investigations and evidence that survives scrutiny.

Source: Code of Practice on disciplinary and grievance procedures (draft) | Acas