Introduction
AI generated grievances are becoming commonplace. Employees are using ChatGPT and similar tools to help them write a grievance, and the document that arrives is longer and more formal than the one you would have received a couple of years ago. It often quotes legislation and references your own policies.
This is not confined to larger employers. Companies with a handful of staff are receiving them too.
The process itself has not changed. A grievance is still a grievance and goes through the same steps it always did. What has changed is how employees write the complaint up. A long, legalistic document may be a serious complaint or a minor one. Its form will not tell you which, and finding out means reading a great deal more text.
This guide covers why these documents run so long, how to find the complaint inside one, and what to do when one arrives.
How Common This Has Become
People Management reported research on this in February 2026. Of the HR directors surveyed, 95% had encountered an employee using AI to raise a grievance or navigate a dispute.
The same research recorded employees using ChatGPT to produce more than ten pages of material for a single grievance.
Smaller employers feel it hardest. A ten-page document raising eight allegations takes the same effort to work through whoever receives it. Most SMEs have no HR department to absorb it.
Why AI Generated Grievances Are So Long
An AI tool does three things to a complaint. It gives it a formal structure, with numbered points, background sections and a summary of what the employee wants. It puts the language into legal terms, so a rota dispute becomes a possible breach of the working time rules. And it expands everything, so two sentences of genuine concern come out as eight pages.
The expansion is the part that creates work for you. The same complaint arrives at five or ten times the length, with the substance spread thinly across it. One incident can appear three times under three different headings, described slightly differently each time. The document then reads as three complaints rather than one.
Why employees reach for these tools
Employees are not necessarily being aggressive when they do this. Many find it hard to put a complaint into words, and the tool gives them confidence that somebody will read it properly.
The practical effect is that the document is a poor guide to what you are dealing with. What sits underneath may be serious or it may be minor, and the length tells you nothing about which. Either way it is the same kind of workplace conflict you have dealt with before.
Whether AI Wrote It Matters Less Than What It Says
Most employers want to know first whether AI wrote the document. The pattern is recognisable, and an employment solicitor writing in People Management has described it in detail. The common signs are American spelling, em dashes in place of ordinary punctuation, and a legalistic tone from an employee who does not normally write that way. The document also runs to pages while saying little about dates and names.
None of it is proof, and the signs are becoming less reliable as AI tools improve at British English.
Knowing the answer also changes nothing you do. You cannot prove it, and you cannot reject a grievance because of it. Raising it with the employee moves the conversation away from the complaint. Recognise the pattern, then set the question aside.
AI generated grievances do tend to appear once you have already started looking into the employee's conduct or performance. Our guide on handling a grievance during a disciplinary covers that situation.
Working Out What an AI Generated Grievance Is Actually Saying
The useful work is reducing the document to the complaint inside it. That is a reading job rather than a legal one, and it takes longer than people expect.
Start with the specific allegations. An allegation has an incident behind it, something that happened on a date and involving named people. Anything describing how the employee felt about the culture is framing rather than an allegation. So is anything about what the law says. A ten-page grievance often contains three or four things that actually happened.
Then group what repeats. Because the tool expands, one incident commonly appears three times over. It shows up in the background section, again under its own heading, and again in the summary. Reading each appearance as a separate complaint is how a four-point grievance turns into a twelve-point investigation.
Take the legal label off
"This may amount to a breach of the implied term of trust and confidence" is a conclusion the tool supplied. It is not something the employee has told you. Take the label off and ask what happened underneath it. The answer is usually a shift change, a comment in a meeting, or a decision the employee thinks was unfair.
Read what the employee is asking for
AI-drafted grievances almost always include a section on the outcome the employee wants. That section is worth reading first, because it tells you what would resolve this. The rest of the document rarely does.
Notice what is not there as well. Length can hide the absence of dates, witnesses and specifics, and you cannot investigate an allegation that has none. Where a point stays vague, ask the employee what they mean rather than guess.
None of this is complicated. It is time, and a long grievance takes a good deal of it.
Why AI Generated Grievances Are Often Inaccurate
AI tools organise a complaint and give it a formal tone. They do not check the facts, and they do not apply employment law reliably.
The same research found that 78% of AI generated grievances relied on inaccurate information or misrepresented the position. Separately, 11% of respondents called the documents full of incorrect information.
The tool has often told the employee what outcome to expect, so they arrive with expectations the situation will not meet.
The claims themselves rarely go anywhere. Of the HR directors who saw an AI-assisted grievance reach an employment tribunal, 86% reported that none of them succeeded. The risk in these cases comes from how the employer handled the process rather than from the document, which is also true of most unfair dismissal claims.
What the New Acas Code Will Expect
Acas is rewriting the Code of Practice that governs grievances, and the draft deals with long written complaints directly.
The consultation on the draft Code opened on 30 July 2026 and closes on 23 September 2026. It is the first full revision since 2009.
Employer and worker representatives both raised concerns about unnecessarily long and complex written grievances. The draft Code therefore encourages employees to give a short, clear explanation of the concern at first, with more detail to follow later. It also asks them to say how they would like the concern resolved, and what they tried informally. Acas has asked separately whether the Code should say anything about the use of AI in these processes.
The draft is not in force yet, and the 2009 Code still applies. That Code matters because tribunals must take it into account. In claims such as unfair dismissal and discrimination, an unreasonable failure to follow it can move an award by up to 25%. That runs in both directions.
What to Do When an AI Generated Grievance Arrives
A grievance does need prompt attention. What it does not need is a quick reply, and the two are easy to confuse. The steps below are what to get right first, before anybody works through the detail.
- Do not reply straight away. Nothing in a grievance needs a same-day answer, and a reply written in the moment tends to cause more problems than it solves.
- Note the date it arrived. Your timescales run from that date, and your policy will say what they are.
- Do not raise the AI with the employee. You cannot prove it, and it moves the conversation away from the complaint you have to deal with.
- Keep it to the people who need it. Forwarding a grievance round the management team creates a confidentiality problem on top of the one you already have.
- Carry on treating the employee normally. Treating someone worse for raising a grievance can breach the trust between you and lead to a constructive dismissal claim. Where the grievance alleges discrimination, it is also victimisation, which is a claim in its own right.
- Get it to whoever handles your HR, the same day. Send the document with a short note on what you think actually happened. The difference between the two is where the useful information sits. With no HR support in place, get advice on your grievance procedure first.
How We Can Help
At Bespoke HR we read AI generated grievances for our clients every week.
Our team reads the grievance and works out what the employee is actually complaining about. We tell you which parts need investigating, which are legal framing with nothing behind them, and what a fair response looks like. Where a formal process is right, we run it with you: the investigation, the meetings, the outcome letter and any appeal.
We also handle disciplinary action where a grievance arrives in the middle of one. Where allegations name a manager, an independent investigation protects both the employee and the decision.
We can also review your grievance policy and put an AI in the workplace policy in place, which is worth having whatever happens with grievances.
Final Thoughts
How employees write complaints has changed. What sits underneath them has not, and neither has the way you handle one.
The work a long document creates is reading work. Somebody has to go through it and find the complaint. That goes faster when they read these every week.
Get in Touch
If a grievance has arrived and you are not sure what to do with it, send it over and we will take a look before you reply.
Frequently Asked Questions
How do I work out what the grievance is actually about?
Read for incidents rather than for arguments. An allegation has something that happened behind it, on a date and involving named people, and the rest is framing. Group the points that describe the same incident, because AI-drafted documents repeat themselves under different headings. Then read the section on what the employee wants, which often says more than the complaint itself.
Do I have to respond to every point in a ten-page grievance?
Every concern needs an answer, though overlapping points often go together as one. Repetition is a common feature of AI-drafted documents, and eight paragraphs often describe the same incident from different angles. Saying in the response how the points group together makes clear that nothing has gone unanswered.
How can I tell if a grievance was written by AI?
There is no certain test, but AI generated grievances follow a recognisable pattern. Look for American spelling and em dashes in place of ordinary punctuation. Watch for a legalistic tone from someone who does not normally write that way. Conditional phrases such as "this may amount to discrimination" are another sign, and the document is often long but vague about dates and names. None of it proves anything, and it does not change how you handle the grievance.
Should I tell the employee I think they used AI?
No. You cannot prove it, the employee will usually deny it, and raising it moves the conversation away from the complaint. It can also look as though you are trying to avoid dealing with what they have said. Deal with the content of the document and leave how it was written out of it.
Can I reject a grievance because an employee used AI to write it?
No. Nothing in the Acas Code says anything about who or what drafted a grievance, only that it is in writing and explains the concern. Refusing to deal with one on that basis does not create a claim by itself. It does expose you, because Acas gives refusing to look into a grievance as an example of what can lead to a constructive dismissal claim. The concerns still need investigating like any other grievance, and AI genuinely helps people who find formal processes difficult.
The grievance quotes laws that seem wrong. What do I do?
Nothing on your own, and do not try to correct it point by point. AI tools frequently attach the wrong legal label to a workplace problem, and some reference legislation and case law that does not exist. Send it to whoever advises you on HR, or get advice if you have nobody. They will separate what the employee is actually complaining about from the legal wrapping around it.
Can I insist on a face-to-face grievance meeting?
You can expect a meeting, but not that it happens in person. The Acas Code puts holding one on you rather than the employee, so a grievance should not be handled entirely on paper. There is a practical reason too, because the employee's right to be accompanied only applies once you invite them to a meeting. Where the employee is disabled you have a duty to adjust the process. That might mean a video call, an interpreter, a support worker or extra time.
Should we have a policy on employees using AI at work?
Yes, and for wider reasons than grievances. A workplace AI policy covers what staff may put into a public chatbot, and an employee pasting colleagues' personal data into one is a data protection issue on its own. A policy also gives you something to point to when AI use causes a problem, rather than making the rules up after the fact.