Artificial intelligence is appearing in an increasing number of HR processes, often without the company being fully aware of it. One employee uses AI to analyse an Excel spreadsheet, another has it summarise a document, while a third wants to compare pay data. However, solutions that make everyday work easier can raise serious questions about data protection and trust, particularly when the organisation has not yet established clear rules for the use of AI. This issue becomes particularly relevant when preparing for pay transparency.

When AI is already present, but the company is not yet aware of it

In many organisations, the use of artificial intelligence is spreading faster than internal policies can keep pace with. Employees often try out various AI tools on their own initiative to simplify their tasks, work faster or find their way more easily through large amounts of information.

This phenomenon is known as shadow AI: when employees use artificial intelligence tools for work that have not been approved by the company or operate outside its IT and data protection controls.

This is particularly sensitive in HR, as a significant proportion of the information handled in this area consists of personal or confidential data. Uploading salary surveys, performance appraisals or spreadsheets containing employee data to an external AI application can initiate a data processing operation whose conditions and consequences may not be fully understood by the employee or even the employer.

The risk increases further if the company has no consistent position on which tools may be used, what information may be shared with them and how AI-generated outputs can be verified.

Employee trust is also at stake alongside data security

Data security risks are usually the first concerns raised in connection with shadow AI. It is important to know where uploaded information is processed, who can access it, how long it is stored and whether it may be used to improve the provider’s systems.

In HR, the purpose for which data uploaded to an AI tool is used and the decisions made on that basis further complicate the situation, as pay analyses or performance appraisal summaries produced in this way may contain inaccurate conclusions which, without appropriate human review, could influence decisions affecting employees.

The introduction of pay transparency makes this issue even more significant. One of the fundamental aims of the new regulation is to help employees better understand the criteria behind pay decisions and gain access to objective, verifiable information. However, if it is unclear what technology the company uses to analyse pay data and how it ensures data protection, this can easily undermine trust in the system. Employees also consider it important to know who uses information about them and for what purpose.

Why is simply banning the use of AI not enough?

Banning the use of unapproved AI tools may be an understandable response from companies, but on its own it rarely provides a comprehensive solution. If employees have already experienced the benefits of the technology and the organisation does not provide a suitable alternative, there is a risk that its use will continue outside corporate controls. An effective approach can therefore be built on four complementary areas:

• Clear policies: the company should define which AI tools may be used, for which tasks and under what data processing conditions.
• Employee education: HR and IT should work together to help colleagues recognise risks relating to data protection, information security and professional practice.
• A secure technological environment: the organisation should provide approved tools that meet corporate data processing requirements.
• Human oversight and responsibility: clear review and approval processes should be established for the use of AI-generated analyses and recommendations.

It is particularly important that employees can understand and apply the policies, as policies that are too general or rely exclusively on prohibitions offer little help in everyday decision-making situations.

Pay transparency takes the issue of AI use to a new level

To comply with the EU Pay Transparency Directive, companies need to manage job, remuneration and performance data in an increasingly structured way. Job evaluation, the development of pay bands, the analysis of market benchmarks and the examination of gender pay gaps can involve significant volumes of data.

AI can indeed provide useful assistance in these processes: it can support the organisation of data, the identification of differences or the analysis of different scenarios. At the same time, using the technology also brings new responsibilities for HR.

When analysing pay data, for example, it is particularly important for the company to verify the methodology used by AI, the accuracy of the data and the soundness of the conclusions. Potential biases may reinforce precisely those inequalities that pay transparency is intended to identify and address.

Establishing rules for the use of AI therefore now extends beyond IT and data protection, as HR leaders must also play an active role in determining the processes in which the technology may be used and under what conditions.

The first step: making existing practices visible

It is worth starting preparations by mapping which AI tools employees currently use and for which tasks. This can help identify the most significant risks, as well as the areas where the controlled use of the technology can deliver real efficiency gains.

The next step is to establish data processing rules, access rights and review processes, and then integrate them into day-to-day HR operations.

The introduction of pay transparency offers companies a good opportunity to take a more deliberate approach to the use of AI as well. Clear rules, an appropriate technological infrastructure and consistent human oversight can together create the conditions for its safe use.