Grievances Written By A Bot: Navigating AI-Drafted Complaints In The Workplace

HR teams across the tech sector are increasingly opening formal grievances that bear the hallmarks of AI generation — structurally formulaic, legally framed, and often far longer than the underlying complaint warrants. Some read as though drafted by a solicitor; others are obviously the product of a hurried AI prompt. What they have in common is that responding to an AI generated grievance presents a distinct set of challenges that demand a thoughtful, proportionate response.

The problem with AI-drafted grievances

On the surface, a well-structured grievance might seem like a good thing.  It should, in theory, be easier to investigate. In practice, however, AI-generated grievances frequently create significant complications.

• Volume and complexity. AI generated grievances tend to be long. For HR teams that are already stretched, a 20-page AI-generated grievance covering multiple alleged breaches requires considerably more resource to investigate properly than a focused, factual complaint, even if the underlying issue is relatively straightforward. HR teams are increasingly finding that they are having to spend a significant amount of time trawling through AI-generated documents which can often be repetitive, trying to identify the key issues/complaints.
• Scope creep and over-inclusion. AI tools, when prompted with a general complaint, tend to generate exhaustive, catch-all documents. They rarely discern between the key events and issues and more minor facts. As a result, it can be much harder to identify the real complaint at the heart of the grievance.
• Disconnection between the written complaint and the underlying facts. Because AI drafts from prompts rather than lived experience, the written grievance may use language and legal framing that the employee themselves does not fully understand or endorse. In some cases, the allegations may bear little or no relationship to the employee’s actual lived experience. This creates a disconnect that can derail investigation meetings, where the employee is unable to elaborate on or substantiate the specific allegations as written. AI tools may also include false or misleading assertions in grievances to make them more compelling, which could result in muddying the issues and making it more difficult to understand the crux of the complaint and the employee’s concerns.
• Inflated expectations. Employees who receive an AI-generated document that reads like a legal letter may develop unrealistic expectations about outcomes, making resolution and settlement more difficult. The language used by AI tools when asked to produce something that will amount to a formal complaint / grievance is likely to be quite aggressive and frame even simple issues as legal wrongs, leading to an unnecessary escalation of matters which could be resolved quite easily. A grievance about a single incident of perceived unfair treatment may arrive citing discrimination, constructive dismissal, whistleblowing, breach of contract and failure to make reasonable adjustments — all in one document and by reference to seemingly unconnected events spanning a long period of time.
• Confidentiality and data protection. Employees may upload confidential information, such as their employment contract, the employer’s internal policies and other people’s data (including sensitive personal data) into AI tools when prompting it to prepare their grievance.  This can be problematic from a confidentiality and data protection perspective if an open AI tool (as opposed to a closed AI tool provided by the employer) is used, and depending on the information uploaded, could constitute a breach of the employee’s own confidentiality obligations.      

Employers’ obligations

Before considering how to manage AI-drafted grievances efficiently, it is essential to be clear about what the law requires, which has not changed. The ACAS Code of Practice on Disciplinary and Grievance Procedures sets out the minimum standards employers must follow and is accompanied by more detailed guidance.

The core obligations are well established: the employer must investigate the grievance, invite the employee to a formal meeting, allow them to be accompanied at that meeting, communicate a decision in writing and offer a right of appeal.  Each of these steps should be taken without unreasonable delay.

The Code does not permit an employer to dismiss or truncate a grievance simply because it is poorly evidenced or overly broad. Every allegation that is reasonably capable of investigation must be investigated.  Employment tribunals are required to take the ACAS Code into account and can uplift any compensation award by up to 25% where an employer has unreasonably failed to comply with it. A failure to appropriately deal with a grievance could also give rise to a standalone claim for discrimination, victimisation or constructive dismissal, for example.

Practical tips for employers

1. Focus on the substance of the grievance
   The fact that AI was used does not diminish the employee’s right to have their grievance heard. Employers should focus on the substance, not the authorship and do not be tempted to disregard a grievance that has been prepared with the use of AI.
2. Use the initial grievance meeting to scope and clarify
   The ACAS guidance encourages employers to hold an initial meeting to understand the nature of the complaint before investigating matters further if necessary.  This initial meeting is critically important in cases where the grievance is AI generated.Where a grievance contains multiple allegations, these should be categorised.  Some may be substantive and require full investigation; others may be speculative, legally misconceived, or entirely unsupported by any facts the employee can articulate.Employers should use this initial meeting purposefully and ask the employee, in plain terms, to explain in their own words what happened, what they are most concerned about, and what outcome they are seeking. This conversation will often reveal that the core complaint is narrower than the written document suggests, allowing the employer to focus its investigation and response accordingly.  This scoping discussion should be documented and shared with the employee so that all parties are clear on exactly what will be investigated and responded to.
3. Keep records
   Given that AI-drafted grievances often contain discrimination or whistleblowing allegations, the risk of subsequent tribunal proceedings is elevated. Meticulous records of every step — the investigation plan, interviews conducted, evidence reviewed and the rationale for the employer’s conclusions — are essential. If the matter proceeds to tribunal, the employer will need to demonstrate that its process was fair, thorough, and proportionate.
4. Consider updating the grievance policy
   The employer’s grievance procedure should be clear that employees are encouraged to raise concerns in their own words.  It is also advisable for policies to include a statement that the employer will seek to understand the employee’s core concern and investigate accordingly, regardless of how the grievance is framed. This would give the employer a documented basis for the scoping approach described above.
5. Train managers
   Line managers may be the first point of contact for a grievance, particularly in smaller organisations.  Employers should therefore ensure that their managers understand that they should not be intimidated by or dismiss an AI-drafted grievance.  The key point is that the employer’s grievance policy continues to apply and that the principles around the standard of investigation and response (i.e. reasonableness and proportionality) remains the same.  A lengthy or aggressive grievance is not necessarily a stronger grievance, but nor is it one that can be ignored.
6. Beware of HR using AI in the response
   AI is available to everyone and there may be a temptation to respond to an AI-generated grievance with an AI-generated response following the relevant meeting(s) and investigation(s).  However, the issues highlighted above regarding AI-generated grievances apply equally, if not more, to AI-generated responses.  There is also a difference between using AI to support the grievance process (collating documents, for example) and to draft the outcome. An employee is likely to know that they are reading something generated by AI from its language and tone and that may make them less receptive to the outcome or believe in its authenticity.  The decision maker may also have to justify their decision (worst case, in the Employment Tribunal as a witness), as communicated in the generated outcome, so it is important that they have appropriately applied their mind to the decision that is in their name.

 
A final thought

AI is not going away, and neither is its use by employees navigating workplace disputes. The employers who manage this trend most effectively will be those who respond to the substance of what their people are telling them, rather than becoming distracted by the sophistication of the packaging or drowning in the detail and legalese.  A grievance drafted by an AI tool is still, at its heart, a signal that an employee is unhappy. Addressing that underlying concern — promptly, fairly, and in accordance with the ACAS Code and established principles of employment law — remains both the legal requirement and, ultimately, the most commercially sensible course of action.