• Nicky

The hidden risks of relying on AI for HR advice

AI tools are becoming increasingly common in workplaces.

Many employers are now using AI not just to draft documents, but to help answer questions such as:

  • Can we dismiss this employee?

  • Can we refuse this flexible working request?

  • Can we extend probation?

  • Does this situation amount to misconduct?

  • Can we make this role redundant?

  • What should we say in a grievance outcome?

On the surface, this can seem efficient and cost effective.

The problem is that AI can provide answers that sound confident, professional and convincing without fully understanding the legal, practical and commercial risks involved in the situation.

That can create a dangerous sense of reassurance.

AI only knows what it is told.

One of the biggest issues with relying on AI for HR advice is that it only works with the information it is given.

In many employee situations, the details that matter most are often the details an employer may not realise are important.

For example:

  • Has the employee recently raised concerns or submitted a grievance?

  • Could there be a disability or underlying health issue?

  • Has there been inconsistent treatment compared to other employees?

  • Is there enough evidence to support the decision?

  • Has the business followed its own procedures?

  • Are there whistleblowing or discrimination risks?

  • How long has the employee worked for the business?

These points can fundamentally change the level of risk involved.

An experienced HR adviser will often identify issues that the employer has not even considered.

AI generally answers the question it has been asked; it does not necessarily identify the questions that should also have been asked.

The process matters as much as the outcome.

One of the biggest misconceptions is that if the wording sounds professional, the approach must be correct.

That is not how employment tribunals work.

A dismissal letter may appear well written, but if the investigation was inadequate, the process inconsistent, or the decision unreasonable, the business may still face significant risk.

AI can help generate wording. It cannot assess fairness, reasonableness, credibility or how a tribunal may view the overall handling of the situation.

That requires judgment and experience.

HR situations are rarely straightforward

Employment issues are often far more nuanced than they first appear.

  • A manager may ask AI for help managing poor performance without recognising that the employee’s conduct could be linked to stress, anxiety, ADHD, menopause symptoms, medication or another health condition.

  • An employer may ask whether they can dismiss an employee during probation without appreciating that certain claims do not require two years’ service.

  • A redundancy process may appear straightforward until consultation; selection criteria or alternative employment issues arise.

The risk is often not the wording itself; it is failing to recognise the wider implications of the situation.

False confidence can create bigger problems

One of the most significant dangers with AI is that it can make employers feel they have taken proper advice when in reality they have only generated information.

Those are not the same thing.

We are already seeing AI generated HR letters and advice that:

  • refer to incorrect procedures;

  • miss important legal considerations;

  • create inconsistencies with company policies;

  • or fail to identify discrimination and employee rights risks.

The output may sound authoritative while the underlying approach remains flawed.

That can become very expensive later.

Confidentiality and data protection also matter

Employers should also think carefully about what information is being entered into AI tools.

HR situations often involve sensitive personal information: health concerns, grievances, disciplinary allegations, performance issues and sometimes medical information.

Entering employee details into public AI systems without fully understanding how that information may be stored or processed can create confidentiality and data protection concerns.

The ICO has already made clear that organisations using generative AI still need to comply with their data protection obligations and handle personal data responsibly.

AI can still be useful.

This is not about saying employers should never use AI.

Used carefully, AI can be a useful administrative and drafting tool. It may help create first drafts of letters, policies, meeting notes or communications.

But employee relations issues are rarely just administrative exercises. They involve risk assessment, judgment, context and experience.

That is where professional HR advice still matters.

A practical point to take away

Before relying on AI generated HR advice, ask yourself:

  • Have we fully understood the risks in this situation?

  • Could there be employment law issues we have not recognised?

  • Are we confident the process itself is fair and defensible?

  • Have we considered consistency and precedent?

  • Would we be comfortable defending this approach in a tribunal?

If there is uncertainty around any of those points, it is worth taking advice before action is taken.

Often the earlier advice is taken, the more options there are available.

How we help

At Rely, we are not against businesses using AI. Used properly, it can be a helpful tool.

But when employee issues arise, the biggest risks are often the things employers do not realise they need to consider.

We help businesses sense check situations before decisions are made; whether that involves a disciplinary issue, grievance, redundancy situation, probation concern or employee complaint.

A short conversation early on can often prevent a much bigger and more expensive problem later.