Generative AI is no longer just a tool for employers. Employees are increasingly turning to tools like ChatGPT to draft grievance letters, prepare disciplinary appeals, compose flexible working requests and formulate arguments for tribunal claims.
Most organisations have thought carefully about how they deploy AI in HR, but less attention has potentially been given to the other side of the equation – employees using AI to engage with HR processes.
In this article, we look at the risks that arise from that trend and set out practical steps employers can take in response.
How employees are using AI in grievance and disciplinary processes
Generative AI tools have become a familiar resource for employees navigating formal HR procedures. An employee facing a disciplinary hearing might use ChatGPT or other AI Tools to draft a detailed response to the allegations against them, complete with legal terminology and references to the ACAS Code of Practice.
An employee raising a grievance might prompt an AI tool to produce a lengthy complaint citing constructive dismissal, the implied term of mutual trust and confidence, or obligations under the Equality Act 2010. This can also result in tactical recommendations such as raising a DSAR or starting without prejudice discussions, which employees might otherwise not be considering at that time.
The result is that correspondence which might previously have been written by a solicitor or trade union representative, or simply by the employee in plain language, is now being produced with a veneer of legal authority. That presents genuine challenges for HR professionals, line managers and in-house legal teams who need to assess and respond to these communications.
Legal and practical risks of employees using AI in HR processes
The increasing use of AI by employees in HR processes creates a range of practical and legal challenges that employers will need to navigate carefully in day-to-day workplace situations.
Risk of inaccurate legal arguments in AI-generated complaint
AI-generated correspondence can contain inaccurate statements of law or misapply legal concepts to the employee’s circumstances. Quite often, employees will take the results and drafting presented by the AI tool and will not check the sources or authorities relied upon.
Generative AI models do not understand the facts of a particular workplace dispute; they generate text based on patterns in their training data. The result can be grievance letters that cite irrelevant legislation, mis-state the legal tests for discrimination or unfair dismissal, or present facts in a misleading way.
For HR professionals without legal training, distinguishing between a meritorious argument and AI-generated content that merely appears authoritative can be a real challenge.
How AI can escalate workplace disputes
When prompted to draft formal complaints, AI tools tend to produce correspondence that is legalistic, assertive and at times, adversarial. This can turn what might otherwise be a straightforward workplace concern into what feels like a multi-faceted legal dispute, making early and informal resolution considerably more difficult.
This matters, given the emphasis that employment tribunals and the ACAS Code place on resolving issues informally and at an early stage. It is important not to ignore AI generated correspondence, and employers should seek legal advice sooner rather than later.
The risk of fake legal references and precedents
Generative AI has a well-documented tendency to “hallucinate,” producing plausible sounding but entirely fictitious information.
In practice, this means AI tools can invent case law references, fabricate statutory provisions, or cite regulatory guidance that does not exist. If an employee’s grievance or appeal references fabricated authorities, you may spend considerable time attempting to verify those references.
Reliance on fabricated material can also undermine the credibility of the complaint and complicate your obligation to deal with it fairly.
Data protection risks and confidentiality concerns when using AI tools
When employees input details of workplace disputes into AI tools, they may inadvertently disclose confidential or sensitive information. This could include personal data about colleagues, details of ongoing investigations, commercially sensitive material, or information subject to legal professional privilege.
Most publicly available AI tools process data on external servers, and many platforms permit the use of inputted data for model training. That raises significant concerns under the UK GDPR and the Data Protection Act 2018, particularly where special category data is involved (for example, information relating to health, trade union membership or protected characteristics).
Workplace policies and procedures need to account for this risk appropriately to help protect the business.
Undermining the integrity of formal processes
Grievance and disciplinary procedures depend on the authenticity of the representations made by the parties.
Where AI is used to draft submissions, there is a real risk that the correspondence does not accurately reflect the employee’s genuine concerns or desired outcome. This can make it harder for you to identify the core issues in dispute and reach a fair decision. This is often shown up during formal meetings, where the employee is unable to articulate the points in their letters, or to point to supporting evidence for concerns they have included.
AI tends to be agreeable
One of the less obvious characteristics (and risks) of AI-generated content is that AI tools tend to be agreeable. They generally accept the user’s framing of events at face value and rarely challenge factual assertions, particularly less advanced models. This means that when an employee inputs an account that is inaccurate or one-sided, the AI is likely to build on it without questioning it.
The result can be grievance letters or tribunal arguments that overstate the strength of a claim, present weak facts as though they are conclusive, and frame modest concerns as serious legal breaches.
This is important for employers to understand, because it can lead employees to believe their case is stronger than it actually is, which in turn can make early resolution and settlement negotiations considerably harder.
How to identify AI-generated grievance or disciplinary submissions
In practice, HR teams and managers may start to notice patterns that suggest AI has been used in employee submissions.
Before looking at how to respond, it is worth thinking about how to recognise AI-generated content in the first place. Common indicators that AI may have been used include:
Unusually polished or legalistic language
The language may be noticeably polished, formal, or legalistic in a way that does not match the employee’s usual style of communication. AI-generated text tends to favour complete, grammatically perfect sentences and rarely includes the typos, shorthand, or colloquialisms typical of most people’s writing.
If an employee who normally communicates in short, informal emails suddenly submits a multi-page grievance letter, that contrast is worth noting.
Repetitive sentence structure and tone
AI tools also tend to produce text with a repetitive rhythm. Sentences are often of similar length, paragraphs are neatly balanced, and the same transitional phrases (such as “moreover,” “furthermore,” or “it is important to note”) appear with unusual frequency.
Superficial or inaccurate legal references
AI-generated grievances may cite legislation or legal principles correctly at a surface level but without connecting them meaningfully to the employee’s specific circumstances.
References to case law should be treated with caution, as generative AI tools are known to fabricate case names, citations, and even the holdings of judgments that do not exist.
Lack of factual or emotional detail
AI-generated correspondence also tends to lack emotional specificity. It may describe a workplace issue in abstract terms without the kind of concrete, situation-specific detail you would expect from someone who experienced it first-hand.
Where a submission is well-constructed but oddly detached from the facts, that can be a telling indicator.
Unusual formatting and structure
AI-generated text frequently uses dashes, bold formatting, and multiple layers of headings and sub-headings in a way that is unusual in everyday workplace correspondence.
If a grievance letter arrives with the visual structure of a published article rather than a personal communication, that is worth noting.
American English spelling or inconsistent tone
Many AI tools default to American English spellings and phrasing (for example, “labor” rather than “labour,” or “utilize” rather than “use”). The tone may also be completely out of character for the individual concerned.
No single indicator is conclusive on its own, and the involvement of AI does not mean the substance of a complaint should be disregarded. However, being able to recognise these patterns will help you engage more effectively with what the employee is seeking to raise.
How employers should respond
Develop a clear AI use policy
Consider developing or updating your policies to address the use of AI tools in workplace processes. A blanket ban on AI is unlikely to be practical or enforceable, but a policy can set out your organisation’s expectations clearly.
For example, it might require employees to confirm that any submissions made in formal HR processes are their own work and accurately reflect their understanding of the relevant facts. It could also highlight the risks of inputting confidential or personal data into third-party AI platforms.
Train HR professionals and managers
Your HR team and line managers should receive training on how to identify and respond to AI-generated content (drawing on the indicators discussed above). More importantly, training should focus on how to respond constructively.
Rather than dismissing AI-assisted correspondence, HR professionals should engage with the substance of the employee’s concerns, asking clarifying questions to understand what the employee genuinely wishes to raise and what outcome they are looking for.
Focus on substance over form
The form of an employee’s communication should not distract from its substance. An AI-generated grievance may be poorly tailored to the situation, but it may nonetheless raise legitimate concerns that warrant investigation.
Equally, a highly polished letter should not be treated as more meritorious simply because it reads well. The focus should remain on the underlying facts and the merits of the issues raised.
Engage in informal resolution where possible
Where AI-generated correspondence has escalated the tone of a dispute, it is worth considering whether informal resolution remains achievable. A conversation with the employee may well reveal that the issues are less complex or adversarial than the written submission suggests.
The ACAS Code encourages early and informal resolution, and that approach should not be abandoned simply because the initial communication reads like a letter before action.
Address data protection risks proactively
Remind employees of their confidentiality and data protection obligations, particularly when formal HR processes are underway. This could be done through existing policies, data protection training, or specific guidance issued at the start of a formal procedure.
Where there is reason to believe that personal data has been disclosed to a third-party AI platform, consider whether a data breach notification is required and take appropriate steps to mitigate any harm.
Review and update contractual provisions
It is also worth reviewing your contracts of employment, staff handbooks and related policies to make sure that confidentiality and data handling provisions are broad enough to cover the use of AI tools.
Consider whether restrictive covenants, intellectual property clauses and confidentiality obligations adequately address the risk of sensitive information being disclosed through these platforms.
Seek specialist employment law advice
Because AI tools are so widely accessible, there is a real prospect that employers will receive a higher volume of formal complaints and tribunal claims, including those that are weak or frivolous in nature. Employees who might previously have been deterred by the difficulty of articulating a legal claim may now feel empowered by AI-generated correspondence that presents their position in apparently strong terms.
Specialist employment lawyers can help you assess the merits of a claim quickly, respond proportionately, and avoid being drawn into protracted disputes where the underlying complaint lacks substance.
What to do next
The use of AI in HR processes by employees is likely to become a routine part of workplace disputes. Employers who take proactive steps now will be better placed to manage risk, maintain fair processes and avoid unnecessary escalation.
The use of AI by employees is not, in itself, objectionable. Employees have always been entitled to seek assistance in preparing formal submissions, whether from colleagues, trade union representatives or solicitors. AI is simply a new form of that assistance.
Reviewing your AI policies, training managers to recognise AI-generated content and addressing data protection risks early can make a significant difference in how effectively issues are resolved.
If you are starting to see more complex or legalistic grievance or disciplinary submissions, early advice can help you assess the strength of the issues raised and respond proportionately.