OUR PUBLICATIONS

If Salaries Are Determined by the Same Algorithm: Competition Law Liability in Price and Wage Setting


05 November 2026 - İrem Gönül

If Salaries Are Determined by the Same Algorithm: Competition Law Liability in Price and Wage Setting

What happens if artificial intelligence determines our salaries? An employer’s use of software to obtain rational wage proposals does not, in and of itself, constitute a violation of competition law. However, if the non-public wage and raise strategies of competing employers sharing the same labor pool are reflected in offers through a common data provider, competition law liability will inevitably come into play. In this scenario, questions will arise as to whether the options presented to employees are truly formed independently in the market.

A similar risk arises when competing sellers set product prices based on recommendations from the same AI model. Therefore, the legal liability of both the companies and the software provider must be dynamically assessed based on the nature of the data fed into the algorithm, the system’s actual operation, and the extent to which it eliminates uncertainty among competitors.

 

The Legal Framework for Price and Fee Coordination

Article 4 of Law No. 4054 on the Protection of Competition explicitly prohibits agreements and concerted practices among undertakings that aim to restrict competition or that have or may have such an effect. The direct or indirect fixing of purchase or sale prices is one of the clearest examples of this prohibition and constitutes an absolute violation. Consequently, the fact that such coordination is carried out via digital software or algorithms rather than traditional methods does not alter its de facto unlawful nature, nor does it exclude the conduct from the scope of Article 4.

The Competition Authority’s Guide on Competition Violations in Labor Markets also addresses wage-fixing agreements precisely within this legal framework. A key point that must not be overlooked here is that undertakings competing in the labor market do not necessarily have to be competitors in the product or service market. For example, two separate companies operating in completely different sectors and having no connection in product markets may compete directly in the labor market to hire software developers with the same qualifications.

 

Shared Software and Indirect Knowledge Exchange

Not all automated pricing tools in the market incorporate artificial intelligence architectures with autonomous learning capabilities. Therefore, during the legal review process, the operational mechanism and technical infrastructure of the technology used must first be clearly established; the phenomenon of algorithmic personalized pricing must be distinctly differentiated from the risk of coordination among competitors. While an undertaking’s practice of analyzing consumer profiles to offer dynamic or personalized prices may be examined under consumer law or the abuse of a dominant position, in a scenario involving the use of a common software provider, the focus is on the anti-competitive nature of the horizontal relationship between competing undertakings.

The use of the same software infrastructure and actual participation in coordination among competitors are entirely distinct phenomena. A company receiving recommendations from the system based solely on its own internal cost data and publicly available sources yields different legal consequences than a scenario in which competitors’ non-public, sensitive commercial data is pooled into a common database. In an investigation conducted along these lines, the timeliness of the data, its level of detail (granularity), and the extent to which it enables the prediction of competitors’ future strategic decisions are of critical importance. Indeed, if competitors have tacitly or explicitly agreed to apply a common pricing methodology or rule, the fact that the inputs fed into the system are publicly available does not automatically render this mechanism lawful.

In these structures, referred to as “hub-and-spoke” in competition law, a common software provider may function as a hub facilitating the indirect flow of information among competitors. The fact that undertakings do not directly view each other’s raw data on their user screens does not, by itself, preclude a legal investigation. It must be thoroughly investigated whether this strategic information belonging to competitors is blended in the background during the algorithm’s recommendation-generation process and what its actual impact is on the companies’ freedom to make independent decisions. Therefore, in an audit, it is not sufficient to merely examine superficial data transfer screens; it must also be technically explained which hidden inputs and optimization objectives the software uses to generate the final recommendation.

On the other hand, with regard to traditional compensation benchmarking activities, Paragraph 28 of the Labor Guidelines provides a cumulative test for enterprises seeking to protect themselves from investigations and remain in a safe harbor. Within this framework, the following conditions must be met collectively: the activity must be conducted by an independent third party; the data source or individual data content must be completely unidentifiable; the data must be at least three months old (historical); there must be at least ten participants in the pool; and no single participant’s data weight may exceed 25 percent. However, this exceptional provision in the guidelines does not provide a priori assurance of compliance or a general exemption for all algorithmic functions and dynamic recommendations offered by a shared software platform.

Finally, the widespread use of a shared software platform in the relevant market (market coverage) may constitute an important presumption in assessing the risk of coordination and the ecosystem’s effects on the market; however, this high usage rate alone does not prove the existence of a violation. Whether the software provider holds a dominant position in its own market and whether it abuses that power through exclusionary or exploitative practices is, however, the subject of a completely separate Article 6 review.

 

Example of Price Setting and Comparison with Product Prices

To illustrate the legal risk posed by algorithmic coordination, let’s consider a hypothetical scenario: Imagine that Companies A, B, and C—all seeking to hire employees with similar qualifications—are using the same salary recommendation service in the market. These companies upload their current salary scales and future raise plans—which have not yet been disclosed to the public—into the system. The software provider, in the background, combines and optimizes this data, recommending that its clients make offers within a very narrow salary range.

In such a scenario, the mere fact that the salary amounts recommended by the system or implemented by the companies are perceived by employees as merely “low” is not, on its own, sufficient to establish a violation of competition law. Any investigation must thoroughly examine whether sensitive, non-public information has been transferred to other employers’ offer structures, the extent of companies’ knowledge regarding the algorithm’s operational mechanisms, and their motivations for participating in the system. If the commercial uncertainty regarding employers’ ability to anticipate their competitors’ moves and thereby make better offers to employees is eliminated, it can be concluded that wage competition in the labor market has weakened.

The parallels and differences between this risk structure in goods and services markets and labor markets are clearly illustrated in the table below: