Since October 5, 2026, companies operating in Spain must explain to their workers how the algorithms and artificial intelligence systems that make decisions about their jobs work. The obligation comes from Royal Decree 723/2026 of September 9, published in the Official State Gazette (BOE) on September 15 and in force twenty days later, which transposes European Directive 2019/1152 on transparent and predictable working conditions.
The transposition is late. As Xataka noted, the directive was about five years behind schedule in being incorporated into Spanish law. The new royal decree implements Article 8.5 of the Workers' Statute, expands the information companies must provide in writing about the essential elements of the contract and fully repeals Royal Decree 1659/1998, which had governed this area for almost three decades.
The most significant change is in letter k) of Article 3.2. It requires employers to report “the existence of algorithmic or automated decision-making systems” when they are used to set or modify working conditions such as the length and distribution of working hours, task allocation, pay, career progression, place of work or termination of the contract.
Notice alone is not enough. The company must also explain “the guidelines, criteria and operating rules” of those systems, so that the worker understands how the different variables affect their working conditions. That information must be provided in writing and before the employment relationship begins, either in the contract or in a separate document, such as the information template that Spain's State Public Employment Service (SEPE) offers on its portal.
According to an analysis by the law firm Cuatrecasas, the rule adds a fourth layer of obligations to an existing framework. The 2021 Rider Law gave works councils the right to know the “parameters, rules and instructions” of algorithms, set out in Article 64.4.d of the Workers' Statute. The General Data Protection Regulation and the EU Artificial Intelligence Act also contain related rights.
The key difference is that the new right is individual. While Article 64.4 of the Statute sets out a collective right for workers' legal representatives, Royal Decree 723/2026 establishes a duty to inform each person. In addition, it is not limited to high-risk AI systems but covers any algorithmic or automated system a company uses to make employment decisions.
The timing also sets it apart from European AI rules. The decree itself relies on several articles of Regulation (EU) 2024/1689 and its Annex III, which classifies certain employment uses as high-risk. However, according to Cuatrecasas, that regulation's employment obligations are postponed until December 2027, whereas the Spanish royal decree applies immediately.
The logic of the rule is clear from well-known examples. Reuters reported that Amazon discovered in 2015 that its experimental recruiting system did not evaluate men and women equally: it penalized résumés containing the word “women's” and downgraded graduates of all-women's colleges. In Italy, the data protection authority fined Foodinho, a Glovo subsidiary, five million euros for unlawfully processing the data of more than 35,000 couriers.
What the rule does not require is just as important. Companies do not have to hand over their algorithms' source code or disclose their models. The obligation concerns guidelines and criteria sufficient for a person to understand what a decision is based on and to challenge it. Jonatan Amenedo, chief executive of Welpay, summed it up in comments to ABC: “Nobody is obliged to hand over their model.”
That boundary between transparency and trade secrecy is likely to be a point of friction. The legal publication Economist & Jurist expects letter k) to generate many queries and suggests that the tension between explaining an algorithm and protecting competitive advantage could end up in the courts. During the passage of the Rider Law, companies such as Glovo had already warned that revealing the details of their algorithms could cost them competitiveness.
For human resources departments, the immediate task is practical: review contract templates against the list in Article 3.2, identify which automated tools play a role in decisions on schedules, tasks, pay or evaluations, and prepare a protocol to answer information requests from existing staff. The obligation applies to employment relationships lasting more than four weeks.
The impact goes beyond delivery platforms. Many companies now use software to plan shifts according to demand, screen applicants, measure productivity or recommend promotions. Under the new rule, any of these uses, if it affects working conditions, must be explained understandably to each employee, which could drive a more careful review of how these tools are designed and audited.
In conclusion, Spain is taking a concrete step toward algorithmic transparency at work, in practice moving ahead of the European AI regulation's timetable. The rule does not fully open the black boxes, but it does require their basic rules to be explained to the people affected by them. Its effectiveness will depend on the quality of those explanations and on oversight by the Labor Inspectorate and the courts.
