AI in HR in Poland – new obligations for employers in recruitment and employee assessment

AI in HR in Poland is increasingly used to support recruitment, application screening, candidate assessment, employee management and the organisation of HR processes. For employers, this means not only new opportunities for automation, but also a growing need to control how artificial intelligence systems influence decisions concerning candidates and employees.

In a previous article, we discussed the broader impact of artificial intelligence on the labour market, skills and the organisation of work: AI and the labour market in Poland – opportunities, risks and challenges for employers. This time, the key issue is more practical: how can employers in Poland use AI in HR processes in a legally compliant way?

The importance of this question is growing due to the European Union’s Artificial Intelligence Act (AI Act). This regulation does not apply only to companies developing AI systems. It may also create obligations for organisations that use such systems, including employers using AI in recruitment, employee evaluation, performance monitoring or HR decision-making. The AI Act follows a risk-based approach, and employment-related AI systems may fall within the high-risk category under Annex III.

For companies operating in Poland, this means moving from a general discussion about innovation to practical compliance management. The implementation of AI in HR should be assessed not only by the HR department, but also by those responsible for employment law, data protection, regulatory compliance, information security and risk management.

Why does AI in HR require particular caution?

HR processes directly affect the professional situation of specific individuals. Decisions made in this area may influence access to employment, working conditions, promotion, task allocation, performance assessment, bonus levels or the termination of cooperation. For this reason, the use of artificial intelligence in HR cannot be treated as a purely technical issue.

The risk is not limited to the possibility that a system may produce an incorrect result. It is also important to understand what data the system uses, which criteria it takes into account, whether the outcome can be understood by a human reviewer and whether it may lead to indirect unequal treatment of candidates or employees.

For example, a system analysing job applications may favour specific keywords, career paths or employment patterns that are not actually necessary for a given position. A performance assessment tool, in turn, may focus too heavily on measurable activity indicators while overlooking work quality, task context or the level of responsibility.

An employer should therefore know not only whether the company uses an AI tool, but also what role that tool plays in the decision-making process. A system that helps draft a job advertisement should be assessed differently from a system that filters candidates, ranks applications or recommends a decision on promotion.

When can an AI system in HR be classified as high-risk?

EU rules on artificial intelligence are based on a risk-based model. The greater the potential impact of an AI system on the rights and situation of individuals, the more extensive the requirements for its use.

In HR, particular attention should be paid to systems that may be classified as high-risk. Annex III to the EU AI Act includes systems used in employment, workers’ management and access to self-employment.

This includes, in particular:

• AI systems intended to be used for recruitment or selection,
• targeted job advertising,
• analysing and filtering job applications,
• evaluating candidates,
• making decisions affecting work-related relationships,
• making decisions affecting promotion or termination,
• allocating tasks based on behaviour or personal characteristics, and
• monitoring or evaluating employee performance and behaviour.

This does not mean that every tool used by an HR department will automatically be a high-risk AI system. The purpose of the system, its function and its real impact on decisions concerning individuals are decisive. The more significant the system’s output is for the situation of a candidate or employee, the more detailed the legal and organisational assessment should be.

Employers in Poland may have their own obligations

A common mistake is to assume that responsibility for AI compliance lies only with the system provider. In practice, an employer using an AI tool may have its own obligations as an organisation deploying the system under its control.

This may be the case, for example, when a company uses an external recruitment platform, an HR management system, a productivity analytics tool, an employee assessment module or a solution supporting HR decisions. Even if the system has been supplied by an external provider, the employer decides in which process it is used, what data is entered into it and how much influence the system’s output has on the final decision.

Before implementing or continuing to use AI in HR, employers in Poland should determine:

• in which HR processes AI systems are used,
• whether the system output influences decisions concerning candidates or employees,
• who supervises the system within the organisation,
• whether users understand the system’s limitations,
• whether candidates and employees are properly informed about the use of AI,
• whether the provider supplies the documentation necessary for compliant use of the tool.

Signing a software agreement with a provider is not enough. The employer should be able to demonstrate that it understands how the tool is used within the organisation and what safeguards have been implemented.

Particular caution is also required where the employer uses a system in a way that differs from its original intended purpose or significantly modifies how it operates. In such cases, the organisation’s scope of responsibility may be broader than in the standard use of a ready-made tool.

For companies that need support in structuring HR processes, employee documentation or payroll-related compliance, professional HR and payroll services in Poland may help ensure that technology is implemented within a clear organisational framework.

AI in recruitment – where does the greatest risk arise?

Recruitment is one of the most common areas in which AI is used in HR. Artificial intelligence systems may support the preparation of job advertisements, CV analysis, initial application screening, candidate organisation, interview question preparation or response assessment.

The greatest risk arises when the system does more than streamline the recruiter’s work and has a real impact on whether a candidate reaches the next stage of recruitment. This may include situations where the tool automatically rejects applications, assigns scores to candidates, creates rankings or recommends the individuals considered best suited to the role.

In such a model, the employer should verify whether the criteria applied by the system are related to the requirements of the position, whether the result can be explained, whether the recruiter can challenge it and whether the candidate is not assessed on the basis of data unrelated to the job.

Particular caution should be exercised with tools that promise to assess personality traits, cultural fit, potential or future performance. Such concepts may be difficult to define precisely, and their automated assessment may lead to non-transparent or discriminatory decisions.

AI in employee assessment and workforce management in Poland

AI in HR may also be used after an employee has been hired. Artificial intelligence systems can analyse work results, activity in company tools, task timeliness, sales data, communication, attendance, team performance or other operational indicators.

Such tools can support organisational management, but they require a careful approach. Employee assessment should not be based solely on an automatically generated result, especially if the system does not take into account the full work context. Quantitative data may be useful, but it does not always reflect quality, task complexity, responsibility level or circumstances affecting work outcomes.

This becomes particularly important when the system’s output may affect bonus levels, periodic appraisal, task allocation, promotion decisions, changes in responsibilities or the termination of cooperation. In such cases, genuine human oversight is necessary.

The person responsible for the decision should be able to verify the system’s output, consider additional circumstances and make a decision that differs from the recommendation produced by the AI tool.

Human oversight and information obligations

One of the key requirements for the use of high-risk AI systems is human oversight. In HR, this is especially important because decisions in this area affect the professional situation of specific individuals. Human oversight should not consist of mechanically approving the system’s output. If a recruiter or manager simply accepts an algorithmic recommendation without a real opportunity to assess it, it is difficult to speak of genuine control.

The person providing oversight should understand what the system is used for, know its limitations, recognise when the output may be incorrect or incomplete, be able to challenge the recommendation and consider information that the system does not take into account.

Transparency towards the people affected by the system is equally important. Candidates and employees should know that an artificial intelligence tool is used in a specific process, especially where its output influences a decision concerning their professional situation. In recruitment processes, it is important to clearly state whether AI is used to analyse applications, filter candidates, create rankings or support assessment. In relation to employees, particular attention should be paid to tools used for monitoring, performance evaluation, task allocation or supporting HR decisions.

AI in HR in Poland and personal data protection

AI in HR very often involves the processing of personal data of candidates and employees. This may include data from CVs, recruitment forms, HR systems, work management tools, performance reviews, business communication or employee activity monitoring systems.

For this reason, the implementation of AI in HR in Poland should be assessed not only from the perspective of the EU AI Act, but also from the perspective of the General Data Protection Regulation (GDPR). The employer should verify, among other things, the legal basis for data processing, the scope of data transferred to the system, compliance with the data minimisation principle, the purpose of processing, the provider’s access to data, the location of processing, information security, the possibility of profiling and the need to carry out a data protection impact assessment.

Particular caution is required in relation to systems that analyse employee behaviour, activity, communication, biometric data or other information that may significantly affect privacy. In such cases, the employer should assess not only the usefulness of the tool, but also whether its use is proportionate.

AI literacy and the regulatory timeline

One obligation that already has practical significance for companies is ensuring an appropriate level of AI literacy. This is not limited to technical knowledge within the IT department. In the case of AI in HR, appropriate preparation should also cover people responsible for recruitment, HR, payroll, employee assessment, compliance, data protection and team management.

Employees using AI systems should understand that the output generated by a tool may be incorrect, incomplete or based on inappropriate assumptions. They should also know what data should not be entered into AI systems, when additional verification is required and how to respond to a result that may lead to an unjustified or unequal assessment of an individual.

Under Article 4 of the AI Act, providers and deployers of AI systems should take measures to ensure a sufficient level of AI literacy among staff and other persons dealing with the operation and use of AI systems on their behalf.

The EU AI Act is applied in stages. Some provisions are already applicable, including rules on AI literacy and prohibited practices. Other obligations are being introduced gradually. According to the European Commission’s AI Act implementation timeline, the regulation is being rolled out progressively, with full application planned in stages.

In the original timeline, rules concerning high-risk systems listed in Annex III were expected to apply from 2 August 2026. However, at EU level, changes connected with the broader digital simplification agenda have been discussed, including possible postponement of certain high-risk AI obligations. Recent reporting on the Commission’s work on high-risk AI classification guidelines indicates that new dates under discussion include 2 December 2027 for certain high-risk areas, including employment, and 2 August 2028 for product-integrated systems.

For employers in Poland, this means that the legislative process should be monitored closely. At the same time, any postponement should not be treated as a reason to delay preparations. Regulatory obligations are not disappearing, and organising HR processes, data flows, documentation and provider relationships requires time.

How should companies prepare for AI in HR processes?

Practical preparations should start with mapping the use of AI in HR processes. This should not be limited only to tools officially described as artificial intelligence systems. AI functions may already be embedded in systems that a company has been using for some time.

This may include recruitment systems, applicant tracking platforms, HR management systems, employee assessment tools, task management systems, analytics tools, activity monitoring solutions and general-purpose generative AI tools used by HR staff.

An AI use map should identify the process in which the system is used, the provider, the data processed, the purpose of use, the person supervising the tool and whether the system output affects decisions concerning an individual. Such an analysis helps distinguish auxiliary tools from solutions that require a deeper legal and organisational assessment.

A separate risk is the informal use of AI by employees. This may include situations where HR staff use publicly available AI tools to summarise CVs, compare candidates, prepare recruitment notes or draft employee evaluations. For this reason, an AI policy should cover both officially implemented systems and the everyday work of HR teams.

Before implementing an AI tool, the employer should also verify the provider. Documentation, transparency of system operation, data processing rules, human oversight mechanisms, error reporting procedures, GDPR compliance and the possibility of control or audit are as important as price and functionality.

Summary

AI in HR is becoming an increasingly important part of recruitment, employee assessment and workforce management in Poland. At the same time, as EU regulation develops, transparency, documentation, human oversight and responsible data management are becoming more important.

Employers should already verify whether their organisation uses AI systems that support decisions concerning candidates or employees. Particular attention should be paid to tools used for application screening, candidate assessment, performance monitoring, task allocation, promotion recommendations or decisions supporting the termination of cooperation.

Even if some deadlines for the application of new obligations are postponed, the regulatory direction remains clear: AI in HR in Poland must be used in a controlled, transparent and rights-compliant manner. Any additional time should be used to organise processes, train staff, verify providers and prepare documentation.

For employers, the key question is therefore not only whether it is worth using AI in HR, but how to do so in a way that is safe, compliant and explainable to candidates, employees and supervisory author