AI pay tools and the EU Pay Transparency Directive

More employers use software to suggest salary ranges, price job offers, rank staff for raises or calculate bonuses. Some of these tools are simple rule engines; others use machine learning trained on past pay data. The EU Pay Transparency Directive does not mention AI, but it sets rules on how pay is decided and explained that apply to whatever tool an employer uses. This page sets out those rules and what they mean for AI pay tools.

The short answer: under Directive (EU) 2023/970, pay must rest on objective, gender-neutral criteria, workers must be able to see the criteria used to set their pay and pay progression, and they can ask for pay information broken down by sex. A tool that cannot explain which criteria drive its pay recommendations makes these duties hard to meet. Member States had to transpose the Directive by 7 June 2026.

Who is covered

The Directive applies to employers in the public and private sectors and to all workers with an employment contract or relationship as defined in each Member State, with consideration to the case law of the Court of Justice (Article 2) (Directive (EU) 2023/970). Its rules on information before employment also apply to job applicants. Pay is defined broadly: the basic or minimum wage or salary and any other consideration, in cash or in kind, including complementary or variable components such as bonuses (Article 3(1)(a)).

Gender-neutral criteria: the core rule for any pay tool

Employers must have pay structures ensuring equal pay for equal work or work of equal value (Article 4(1)). Those structures must allow an assessment of whether workers are in a comparable situation on the basis of objective, gender-neutral criteria, agreed with workers' representatives where they exist. The criteria must not be based directly or indirectly on sex, and must include skills, effort, responsibility and working conditions, and where appropriate other factors relevant to the job. Relevant soft skills must not be undervalued (Article 4(4)).

For an AI tool, this is the hard part. A model trained on past salaries learns the patterns in those salaries, including any past gaps. Inputs that look neutral, such as previous pay, negotiation history or hours of availability, can track sex indirectly. The Directive defines indirect discrimination as an apparently neutral provision, criterion or practice that puts persons of one sex at a particular disadvantage, unless objectively justified by a legitimate aim with appropriate and necessary means (Article 3(1)(j)).

What workers and applicants can see

• Before hiring: applicants have the right to the initial pay or its range, based on objective, gender-neutral criteria, for example in the job advert or before the interview. Employers must not ask applicants about their pay history (Article 5). A tool that prices offers from a candidate's current salary works against this rule.
• Pay setting and progression: employers must make easily accessible to workers the criteria used to determine pay, pay levels and pay progression, and those criteria must be objective and gender neutral. Member States may exempt employers with fewer than 50 workers from the pay progression part (Article 6).
• Right to information: workers can request in writing their individual pay level and the average pay levels, broken down by sex, for workers doing the same work or work of equal value. Employers must answer within two months at the latest, must tell all workers of this right every year, and cannot stop workers from disclosing their own pay to enforce equal pay (Article 7).

If pay is set or recommended by software, these rights in practice reach into the tool: the employer has to be able to say which criteria it applies and to produce the comparisons workers ask for.

Pay gap reporting and joint assessments

Employers with 250 or more workers must report on their gender pay gap by 7 June 2027 and every year after; employers with 150 to 249 workers by 7 June 2027 and every three years; employers with 100 to 149 workers by 7 June 2031 and every three years (Article 9). The report covers the mean and median gap, the gap in complementary or variable components such as bonuses, the share of women and men receiving them, quartile pay bands, and the gap by category of workers. Workers' representatives must be consulted on its accuracy and have access to the methodologies the employer applied (Article 9(6)).

Where the report shows a difference of at least 5% in average pay between women and men in any category of workers, the employer has not justified it on objective, gender-neutral criteria and has not remedied it within six months, the employer must carry out a joint pay assessment with workers' representatives (Article 10).

Who must prove what

When a worker establishes facts from which discrimination may be presumed, it is for the employer to prove there was none. Where the employer has not met its transparency obligations under Articles 5, 6, 7, 9 and 10, the burden is on the employer to prove there was no discrimination, unless the failure was manifestly unintentional and of a minor character (Article 18). An employer that relies on a tool it cannot explain is badly placed to carry that burden.

How the AI Act adds to this

Under the EU AI Act, AI systems intended to make decisions affecting terms of work-related relationships, or to monitor and evaluate performance and behaviour, are listed as high-risk (Regulation (EU) 2024/1689, Annex III, point 4(b)) (AI Act). An AI system that sets pay or ranks staff for raises can fall within that wording, which brings duties such as human oversight and informing workers' representatives before use, applying to employment systems from 2 December 2027. See AI in performance reviews and consulting employees before an AI rollout.

Practical steps

• Employers: list every input the pay tool uses and drop pay history and other inputs that track sex. Write down the criteria in plain words so they can be shown to staff. Test outputs by sex within each category of workers before using them, and keep a person accountable for every final pay decision.
• Workers: ask for the criteria used to set your pay and progression, and if needed for the average pay by sex for your category. Your representatives can ask for the methodology behind the pay gap report.

For the current rules in Great Britain on discussing pay and gender pay gap reporting, see UK pay secrecy and pay gap reporting. National laws that transpose the Directive set the detail, including penalties and how categories of workers are defined, so check the rules in your country. For tools that set pay per task on platforms, see the EU Platform Work Directive and algorithmic pay. This page is general information about EU rules, not legal advice.