Many employers are seeing a trend: grievances, appeals and Employment Tribunal (“ET”) claim documents appear to have been drafted, at least in part, using AI. HR teams and managers can feel caught in a battle of human resources versus artificial intelligence and face lengthy submissions packed with legal arguments, case citations (which are not always accurate) and threats of litigation.

It is no longer simply "computer says no", the computer now says "no" in 3,000 words, cites five authorities and demands a response within four working days.

What are the challenges for employers? 

In many ways, they are obvious but the peculiar challenges of AI are:

• It takes a "kitchen sink" approach, identifying every conceivable claim arising from the facts;

• Straightforward issues get reframed as potential claims, and a grievance that should simply flag workplace difficulties can escalate into serious legal allegations;

• The documents drafted often adopt an adversarial, legalistic tone which can damage working relationships and entrench positions;

• AI text can obscure the employee's actual concerns;

• It can tell the user what they want to hear, reinforcing the employee’s perspective and creating an echo chamber. By repeatedly validating how strong their claim is or how badly they have been treated, AI can inflate expectations; and

• Assumptions and “hallucinations" require checking against the underlying complaint and evidence.

These issues can place a disproportionate burden on employers, requiring HR teams and managers to spend considerable time unpicking lengthy submissions, identifying the employee’s underlying concerns and checking allegations against the evidence. Time and resources that could otherwise be devoted to addressing the problem are instead spent separating genuine issues from repetition, inaccuracies and legal argument. 

What are an employer’s obligations when handling an AI grievance?

Grievances are governed primarily by the ACAS Code of Practice on Disciplinary and Grievance Procedures (the “Code”). Non-compliance is not itself unlawful, but tribunals must consider it and may adjust compensation by up to 25% for either party’s unreasonable failure to comply.

The current Code requires employers to:

1. act promptly and consistently;
2. hold a meeting without unreasonable delay, allowing the employee to explain their grievance and proposed resolution;
3. investigate as necessary to establish the facts;
4. communicate the outcome in writing without unreasonable delay; and
5. offer a right of appeal.

The involvement of AI does not change these basic principles. An employer should resist the temptation either to dismiss a grievance because it appears to have been AI-generated or to treat every assertion within it as requiring a separate, extensive investigation simply because of the way that it is presented. 

The starting point should remain the same, what is the employee complaining about, what are the relevant facts and what needs to be investigated in order to reach a fair conclusion?

Are there challenges for employees? 

Although AI can help employees organise their thoughts and obtain general information about employment claims, many of the concerns outlined above apply equally to them. It is important that employees realise that the AI may be mismanaging their expectations as to outcome and may rely on factual errors (a client of ours received a grievance with references to internal emails which did not exist), misstated law or invented details presented as fact.

If an employee subsequently accepts that parts of their grievance were the product of an AI hallucination, this may inevitably raise questions about the accuracy and credibility of the grievance more broadly.

Confidentiality and Data Processing

Unauthorised public AI tools also pose risks to confidential information, commercially sensitive material, intelectual property and colleagues’ personal data.

An employee is likely to upload internal company emails, documents, performance reviews, client information, personal data relating to colleagues or other commercially sensitive information into a public AI platform. This raises a number of important questions, where is that information being stored? Who has access to it? Has personal data been transferred outside the UK? Our IP team can provide further advice on these issues.

What can/should an employer do now? 

AI is not going away. What can employers do?

• Review grievance procedures and templates; require key allegations, dates, evidence and the desired outcome upfront, and set page or word limits.  
• Use meetings to clarify issues;ask employees to explain their concerns and desired resolution in their own words, and clarify inaccuracies. Hold a preliminary meeting if needed to establish the complaint’s scope.
• Train managers and HR to recognise AI-assisted submissions; ensure that managers are trained to focus on the  substance of the grievance, not length, tone or legal jargon. There is no need to rebut every legal argument in a response to a grievance.
• AI policies; consider introducing a specific AI policy, supported by appropriate training, so that employees understand not only what is prohibited but also why these restrictions matter.
• Maintain clear investigation records; given the volume of material AI can produce, keep a clear audit trail of issues investigated, the evidence that has been considered and conclusions reached.
• Review AI and data protection policies; review your existing policies, including your data protection, confidentiality, IT and acceptable-use policies, to ensure that they robustly address the use of AI tools. These policies should expressly prohibit uploading confidential information, personal data, client information or commercially sensitive material to public AI tools. Repeat these restrictions in the grievance policy and state that breaches may result in disciplinary action.