Picture this: you receive a grievance from an employee. It’s lengthy, formal, and packed with references to employment law and case law. At first glance, it seems well put together – almost too well put together – and something about it doesn’t feel quite right.
Furthermore, the details in the grievance do not align with the reality of the situation, and the employee is refusing to meet in person, insisting that the matter be handled in writing.
If this sounds familiar, you may be dealing with a ‘GIT’ – a grievance invented by technology. It’s just one example of how AI is increasingly being misused in the workplace.
Lawyers are seeing an increasing number of enquiries from clients who suspect their employees may be using AI tools, such as ChatGPT, to draft their grievances. And it doesn’t stop there – they’re also seeing AI used in appeals, responses to emails, and, perhaps more concerningly, in outcome letters prepared by managers.
How employers can spot and respond to complaints crafted by AI
Many of the enquiries lawyers see in relation to questionable grievances follow a similar pattern. The grievance tends to be unusually long, detailed, and formal. It’s often raised in the middle of an investigation into the employee’s conduct or performance. In many cases, the employee is reluctant to attend face-to-face meetings, instead requesting that grievance or disciplinary processes be conducted entirely in writing.
Taken together, these patterns may suggest that the employee has used AI tools such as ChatGPT or Copilot to draft the grievance.
Why AI grievances are a problem
A key issue with employees using AI to draft grievances is that minor incidents can be blown out of proportion. Innocuous actions or behaviours may be presented as bullying, sexual harassment, or discrimination, creating a distorted view of the situation.
AI can take minor incidents and exaggerate, embellish, or re-interpret them in various ways. As a result, when an employer receives a GIT, it’s not always clear which parts reflect what actually happened and which have been ‘enhanced’ by AI. This creates uncertainty, making it harder to assess the true nature of the grievance and respond appropriately.
What makes this even more problematic is the ease and speed with which AI tools can generate paragraphs of detailed, persuasive text. Employees who might not have otherwise taken the time to raise a formal grievance can now spin up convincing complaints in a matter of seconds, increasing the likelihood that grievances are submitted.
The challenge is compounded by the fact that many GITs are submitted during ongoing disciplinary, redundancy, or other sensitive processes. Employees often turn to AI when they feel threatened about their job, which can result in grievances that are highly detailed and, again, inflated. Employers must investigate these curveball grievances – carefully separating fact from embellishment – while simultaneously managing the original process. Naturally, this increases the complexity and sensitivity of the situation.
The difficulty for employers is that, unless there is a clear reason not to, a GIT must be treated as a genuine grievance and investigated accordingly. Even when a GIT is excessively long and includes references to incidents that either didn’t occur or have been overblown, each element still needs to be carefully considered and, where possible, resolved. This, of course, can take up valuable time and resources.
How to spot AI grievances
Given the proliferation of AI, particularly on social media, many lawyers have developed a sense of whether content has been AI-generated. Of course, not everyone is familiar with AI-generated content or skilled at recognising its hallmarks.
With the rise of employees using AI at work, employers will want to look out for telltale signs that text may be generated by AI.
These include American spellings, such as “behavior” instead of “behaviour”, “organization” instead of “organisation”, and “favoritism” instead of “favouritism”, and also the use of American grammar or treating collective nouns as singular (“the team is” vs British English “the team are”).
Then there is the frequent use of dashes (–) instead of standard punctuation, such as commas, colons, or semi-colons; the overuse of stock phrases or transitions, such as “in conclusion”, “it is important to note” or “as mentioned above”; and an overly formal or unnatural tone, particularly from individuals who would not normally communicate that way or an overly balanced or neutral tone, even when the topic might naturally invite opinion or emotion.
Other giveaways are vague or generic content, such as claims lacking specifics, unrealistic examples, or no reference to personal experience or context; conditional or hedging language, such as “this could amount to discrimination” or “this may be a protected disclosure”; and the repetition of ideas or synonyms, where the same point is made multiple times in slightly different words.
Recognising these patterns can help identify when a grievance may have been AI-generated.
Dealing with AI grievances
AI isn’t going away, and employees will continue to use – and sometimes misuse – it. To manage the rise in AI-generated grievances, employers are going to have to trust their instincts by learning the common signs of AI-generated text so they can spot them more easily, and if something feels off, run it through an AI detection tool to confirm or disprove their concerns.
They should compare notes so that if they suspect a grievance has been AI-generated, it should be reviewed alongside other documents or communications from the employee to see whether it’s consistent with their usual language/style.
Expert advice is essential; if there’s uncertainty, the document should be shared with a legal or HR specialist for a second opinion.
It makes sense to review and update the grievance policy to ensure that any grievance requires a face-to-face meeting. However, in some cases, a remote meeting may be a reasonable adjustment.
An employer shouldn’t dismiss a grievance simply because it appears AI-generated. Even if the text has been drafted with AI, the concerns raised may still be genuine. Every grievance should be treated seriously until there is reason to believe otherwise.
Summing up
AI is here to stay, and we’re all going to have to learn how to use it and spot any signs of abuse. One thing is clear, just because a claim has been levelled – with the help of AI, possibly – doesn’t necessarily mean that the claimant is correct in their assertion or may win. Employers will still need to investigate the matter fairly and thoroughly.