AI is changing the way employees raise workplace complaints, with the pace of change faster than employers may realise.
Tools such as ChatGPT are now being used routinely by employees to draft formal grievances, appeal disciplinary decisions, challenge redundancies, and even prepare employment tribunal claims.
The result is a new kind of complaint landing in HR inboxes: polished and often far longer than anything the employee might have written unaided.
Kerrie Hunt, Partner in our Employment team looks at how, that shift carries some real practical consequences.
What's actually happening with AI drafted employee claims
The numbers speak for themselves. According to survey numbers reported by People Management, 83% of HR directors claimed to have worked with businesses facing issues related to employee AI usage.
The survey found that 95% of respondents had dealt with cases where AI had been used to raise grievances or progress disputes.
Of these cases, 78% saw the AI-generated claims use inaccurate information with a further 11% full of incorrect information.
Commonly, submissions carry the hallmarks of AI-generation – formal language, detailed legal references and a length and confidence that perhaps do not bear any similarity to how the employee normally communicates in writing.
The problem is not just the volume. It is the accuracy. AI tools do not verify facts, and they regularly generate legal references that are wrong, irrelevant, or simply made up. In one documented case, a grievance letter cited nine legal cases – only two of which were real. The rest had been fabricated by the AI tool.
Reports have also been made of similar examples involving submissions that cite US case law, out-of-date legislation, or legal principles that have no bearing on the employee's actual situation.
This matters because employers cannot simply dismiss a complaint on the basis that it was written by a machine. Under the Acas code of practice, all grievances must be investigated thoroughly, regardless of how they were prepared.
That obligation does not change if the language sounds unusual, the legal references appear dubious, or the document runs to 10 pages, when a short email might have said the same thing.
Why employees are turning to AI
The appeal is obvious with AI tools being fast, free or low-cost, and readily accessible. For an employee who feels intimidated by formal processes, or who simply does not know where to start, they offer a way to produce a document that sounds authoritative and considered.
That can be a legitimate benefit. Some employees with genuine complaints are better able to articulate their concerns with AI assistance, particularly those for whom written English is not a first language.
But the tools carry real limitations beyond inaccurate legal references. They can also generate unrealistically high estimates of compensation, making early settlement conversations almost impossible to have sensibly.
One further risk is easy to overlook: data protection. When an employee pastes internal correspondence, contractual terms, or personal information about colleagues into a public AI platform, control over that data is effectively lost.
Open AI platforms are not equivalent to legal advisers or licensed enterprise AI products operating within secure environments, and their commercial terms often allow for data retention or use for service improvement. Employers should be alert to this and ensure their own policies address it clearly.
The burden on HR teams
AI-generated grievances are taking longer to read and are harder to evaluate. The challenge falls most heavily on small and mid-sized businesses that lack the resources to work through lengthy submissions filled with incorrect legal references and inflated assertions.
Meanwhile, the tribunal system is already under significant pressure. Government figures show that single claim open caseload at the employment tribunal rose by 32% in 2024/25 compared to the year before, with single claims reaching their highest level since 2012/13 by the end of 2025.
Following its introduction into law last year, the Employment Rights Act 2025 – which includes provisions such as reducing the unfair dismissal qualifying period and extending tribunal time limits – is expected to push volumes higher still, with an influx of AI-generated claims, however inaccurate, only adding to that burden.
What employers can do
The best response is not to try to detect whether AI was used, or to treat AI-assisted complaints as somehow less valid as this can divert attention from the substance of the complaint. What matters is having robust, consistent processes in place.
Some practical steps you can consider:
- Review your grievance and disciplinary policies. Add a short declaration requiring employees to confirm that any submission represents their genuine account of events. This encourages employees to read what they submit and take personal responsibility for it.
- Update your AI use policy. Set out clearly what employees can and cannot use AI tools for in a work context, including the data protection risks of inputting confidential or personal information into public platforms.
- Take a face-to-face approach. Where it is clear AI has been used to generate the grievance, sit down with them for an in-person meeting to cut through and get their account of what happened and how it is impacting them in their own words. Where a grievance is unclear, the Acas code supports inviting the employee to discuss the issues in person before the formal investigation begins.
- Stay focused on the facts. Whether a document was drafted by a human or a machine, the same investigation process applies. Once you have clarified their concerns, stayed focused on investigating promptly. Do not get side-tracked by AI-generated legal assertions or alleged claims nor take them at face value but, instead, investigate the facts and check them.
- Get advice early. Where a grievance includes serious allegations such as discrimination, whistleblowing, or multiple overlapping claims, take legal advice before you respond. The stakes for getting the process wrong are significant, and a tribunal can adjust any award by up to 25% for unreasonable failure to comply with the Acas code.
Employees using AI tools will only increase. The question for employers is not whether to engage with AI-assisted complaints, but how to do so consistently and without unnecessary risk.
It’s important to remember that the fundamentals of good employment practice have not changed. The changes lie in the volume and complexity. Getting processes in order now to future-proof your ability to protect the business is the way to stay ahead.
Thrings’ Employment lawyers are experienced in dealing with business matters that affect the workforce and has acted for both employers and employees from start-ups and SMEs all the way to multinational corporations across a wide range of employment matters. To find out how they can help strengthen your polices, and solve disputes, please get in contact.
