How AI Is Changing Employment Disputes in England and Wales

For years, employment lawyers and HR professionals have grown accustomed to receiving long, detailed, and increasingly complex complaints, grievances, and employee communications. Now, an employee can “up the ante” further still by entering workplace experiences, emails or concerns into an artificial intelligence (“AI”) tool and ask it to identify legal issues, suggest further evidence, draft a grievance, prepare an appeal or explain what remedies are available. AI assistance can extend across the lifecycle of a dispute, from grievances and settlement correspondence to tribunal claims and subject access requests.

AI use may make employees feel empowered by increasing access to justice. But there are several issues with this approach. AI is only as good as its prompts. It relies on the employee’s subjective experience, may miss the context behind decisions and may lack the complete picture needed to provide grounded, balanced advice. AI tools may also misstate the law, misinterpret case authorities, or “hallucinate” cases that do not exist. Employees may consequently become entrenched in a position that amplifies tension with their employer and pursue unrealistic settlement figures.

New Regulatory Focus

In July 2026, Acas launched a consultation on a revised Code of Practice on disciplinary and grievance procedures. The consultation specifically asks whether issues arising from the use of AI in disciplinary and grievance processes should be addressed in the Code or associated guidance. The draft Code also places greater emphasis on early resolution and encourages workers to keep written grievances short and clear.

Workplace Concern Becomes a Full Legal Proposition

Due to AI, an employee who might previously have sent a short email can now produce a carefully structured grievance containing a chronology, legal terminology, multiple alleged breaches, and a list of requested outcomes. That can be helpful. Employees have legitimate interests in understanding their rights and articulating concerns clearly. But there is a distinction between a detailed document and a strong case. AI tools can turn 10 overlapping concerns into 10 apparently distinct allegations. It can convert an employee’s interpretation of an event into a confident legal proposition. For HR, responding to every sentence individually can create an unintended consequence: a grievance which might have been capable of resolution becomes an increasingly technical investigation into dozens of allegations.

The solution here is surprisingly low-tech. Talk to the employee in person and establish:

• What are the underlying concerns?
• Which allegations are genuinely distinct?
• What facts are agreed and what facts are disputed?
• What outcome are they seeking?
• What would resolve the concern?

This is particularly relevant as the proposed Acas Code places greater emphasis on early and informal resolution, including conversations, mediation and facilitated discussions. It will also become increasingly important to “cut through” to determine the key allegations and points of contention.

Spotlighting Previously Unknown Remedies

As AI makes employment law resources more accessible, employees are also pursuing some previously rare and relatively unknown remedies. One example is interim relief. In certain types of automatic unfair dismissal claims, including whistleblowing cases, an employee can apply for interim relief. If interim relief is granted, the tribunal may order the employer to reinstate the claimant or re-engage them in an equivalent role. It may also make a “continuation of contract order,” which has the effect of keeping the employee suspended on full pay until the full hearing. Given that we are now seeing a significant backlog of claims and huge delays in listing claims across employment tribunals, the amount at stake can be significant.

In June 2026, Employment Tribunal Presidential Guidance on applications for interim relief was issued. The Guidance was prompted by an increase of around 1,200% in (i) interim relief applications in the United Kingdom, seemingly AI-assisted, and (ii) the amount of documentation accompanying those applications. The threshold for success in interim relief applications is high: that is, the employee must show a “pretty good chance” of succeeding at a final hearing. Most interim relief applications fail, but employees may regard them as worth the gamble because of the potential upside of receiving full pay until the final hearing, or a significant settlement. If an employee succeeds at the interim relief stage but is later unsuccessful at the final hearing, the employer cannot claw back those payments. Despite the high threshold for success, AI tools appear to have assisted and enabled employees to bring such claims. This may show that while AI can sound persuasive and helpful, it can lack the practical experience or context needed to recognise when the prospects of success are too low to justify pursuing a claim.

AI Reinforcement

The use of AI tools carries a subtler risk: AI is very good at helping users articulate a position. If a user repeatedly presents the same facts on the assumption that discrimination, whistleblowing, or some other wrongdoing has occurred, the resulting analysis can reinforce that framing. An employee can spend days refining one interpretation of events before the employer has the opportunity to explain what happened from its perspective.

This is where prompting becomes important. Compare: "Explain why my employer has discriminated against me," with "Here are the facts as I understand them. Identify the strongest arguments for and against a discrimination claim, the assumptions I may be making, what evidence would be needed and what alternative explanations should be considered." The second prompt is much closer to good legal analysis. There is a lesson for employers too. HR professionals using AI to analyse a grievance should not simply ask "What risks does this create?" They should also ask: what assumptions does this analysis depend on? What evidence would undermine it? What alternative explanations should be investigated? AI literacy should therefore mean more than knowing how to use AI; it should include knowing how to challenge it and use it to challenge yourself.

Confidentiality and Data Protection

Employment disputes can contain highly sensitive information, such as health and disability information, allegations of discrimination, employee names and other personal data, salary information, family circumstances, disciplinary records and commercially confidential material. Yet the practical temptation is obvious. An employee may receive an email from their manager and paste it into an AI tool: "Is this discriminatory?" An HR professional may receive a grievance and ask: "Summarise this and identify the key investigation questions." A manager may want to draft a performance letter and provide an AI tool with details of the employee’s performance and absence history. The convenience is considerable, but so is the potential risk. The ICO’s work on generative AI recognises that personal data included in prompts can engage data protection obligations. A simple instruction saying "do not use AI" is unlikely to be enough. If employees already have access to AI tools, an unrealistic prohibition may simply encourage unapproved use. Organisations should instead decide which AI systems are approved, what information may be entered, and what restrictions apply to sensitive data. They should also consider how AI providers use and retain submitted information, when anonymisation is appropriate, whether a data protection impact assessment is required, how AI use is monitored, and what training employees receive.

AI in the Employment Contract

Employers should consider including express contractual provisions dealing with AI use, particularly for employees who regularly handle confidential information or sensitive personal data. A sensible framework combines contractual obligations with an AI acceptable-use policy covering permitted use, prohibited inputs, verification requirements, human accountability, disclosure obligations, approved systems and monitoring. There is also a cultural question. If employees believe that admitting to AI use will automatically result in disciplinary action, they may simply stop telling the organisation that they use it. This is known as “Shadow AI”. A good policy should aim, where possible, for responsible visibility rather than artificial prohibition.

The Employment Dispute

One of the most significant consequences is what happens when both sides start using AI. The employee has an AI assistant, the employer has an AI assistant, and the employment lawyers have their AI assistants. The grievance, investigation plan, response, and appeal may all be AI-assisted. There is a risk that people stop having a workplace conversation and start exchanging machine-assisted legal positions. Human judgment therefore becomes more, rather than less, important. AI can help identify issues, organise information, and suggest questions; it cannot decide which evidence is reliable, which issues matter, whether somebody’s account is credible, or what outcome is appropriate. While AI can assist the employment process, it should not become the employment process.

What Should HR Do Differently?

The best practical response is to become more deliberate about where human judgment adds value. If you work in HR or employment law, consider the following practical steps:

1. Assume the use of AI is already part of the employment relationship. Build processes on the basis that employees, managers, and advisers may use AI in complaints, investigations and correspondence.
2. Focus on the issue beneath the document. Identify the underlying concern, the facts in dispute, and the outcome the employee is seeking, rather than responding to every AI-generated allegation.
3. Bring employees into the room early. Use conversations, mediation, or facilitated discussions to test assumptions and explore resolution before positions harden.
4. Train people to challenge AI outputs. Ask what assumptions an output depends on, what evidence would undermine it and what alternative explanations should be investigated.
5. Establish practical confidentiality rules. Define which tools are approved, what information may be entered and when anonymisation or additional safeguards are required.
6. Review contracts and policies. Keep AI acceptable-use provisions, confidentiality obligations, verification requirements, and human accountability aligned with actual workplace practice.
7. Look for patterns across grievances. Use recurring concerns to identify organisational issues and address causes, not just individual claims.

AI is transforming employment disputes, but the organisations that respond effectively will be those that invest in human judgment, early resolution and clear governance frameworks. The more easily AI produces sophisticated legal documents, the more important it becomes to keep the human conversation at the centre of the employment relationship.