Inteligencia artificial

This post is also available in: Español

On February 19, 2020, the European Commission published the White Paper on Artificial Intelligence (“AI”), as we explained in this blog post. The Commission subsequently launched a public consultation to receive stakeholders’ opinions and public positions. The preliminary results published by the European Commission are available here.

The Spanish Markets and Competition Commission and the Catalan Competition Authority (“CNMC” and “ACCO,” respectively) made a joint contribution to the EC’s consultation, identifying the main legal challenges posed by AI in an increasingly digital economy and proposing measures and instruments for competition authorities to react to them. The authorities’ report, published on September 30, 2020 is available at this link.

The report identifies the main challenges posed by AI in competition law, notably including:

  • The impact of zero-pricing models, in which the price factor loses significance in the legal analysis by competition authorities;
  • The need for competition authorities to have the capacity to react quickly to address the challenges of the digital environment. This capacity is essential in view of the speed of the digital environment and the acceleration of the network effects on platforms as a result of using AI;
  • So-called data mergers, in which it is often difficult to identify operational efficiencies and anticipate the potentially affected markets;
  • The risk of algorithmic collusion due to competitors increasingly using algorithms and other deep learning tools to agree prices, whether through a coordinated strategy or tacit coordination with no interaction between them, making it harder to identify; 
  • New barriers to entry linked to AI , such as access to data; and
  • The risks associated with hyper-personalization of services, which can lead to other competitors being excluded in favor of specific operators. For example, an assistant voice guide offering a single response rather than a list of results, thus promoting its products or a specific operator with which it has reached an agreement.

To address these challenges, the competition authorities propose specific measures. They mainly suggest adapting the regulations so that competition authorities can use AI and be more open to the knowledge existing in this area:

1. Adaptation of the tools available by authorities to control the use of AI and identify anti-competitive practices such as algorithm-based collusion.

Three cases of collusive practices using algorithms have been sanctioned at European level: in 2016, the British CMA issued a penalty in a price-fixing case relating to online poster sales (accessible here); in 2018, the Commission sanctioned four electronics manufacturers for distributors using price-monitoring algorithms (accessible here) and; in 2019, the British energy market regulator OFGEM fined two energy companies and the software provider and consulting firm for agreeing to distribute markets and customers using software (accessible here). The CNMC has announced that it is investigating a possible case of algorithmic collusion in the real estate brokerage market, allegedly through a real estate management computer program and its algorithms (press release available here).

In these cases, the sanctioned conduct was committed using algorithms and was detected with traditional tools. However, the competition authorities are aware of their limited capacity to detect these cases, conscious that the use of price-fixing algorithms by online distributors is increasing. This could be due to the lack of specialist profiles in this area within the authorities and a lack of staff training.

To adapt the available tools, the report proposes: (i) to have access to the information required to properly check companies’ conduct and detect anti-competitive conduct ex officio. Based on the report, this information includes “that held by the Public Administrations, including information on public procurement, an area particularly affected by collusive conducts, but also information on companies that would make it possible to monitor their conduct, particularly in pricing (such as, for example, data collected through cookies) and even information on users (to determine, for example, whether they are subject to discrimination)”; (ii) to be able to supervise the code of the algorithm used and access information available in computerized or electronic form, databases, applications, computer services, digital platforms and application programing interfaces as part of an investigation. In the authorities’ opinion, the ECN+ Directive offers Member States an opportunity to strengthen investigatory powers in this area; and (iii) to study the markets that use software, hardware and cloud computing to implement AI techniques in view of the barriers to entry they may arise. To guarantee competition in the AI services market, authorities believe it is necessary to ensure that operators can access that processing capacity neutrally.

2. Increase openness of competition authorities to acquire knowledge on how AI systems work with (i) other bodies or public administrations that are using these tools, (ii) research centers and universities, and (iii) data protection authorities to obtain external knowledge about the use of AI. They propose to cooperate with these bodies to create synergies and generate new ideas and applications for AI techniques.

After a thorough analysis of the multiple responses received in the public consultation, the European Commission will submit a regulatory proposal. We will monitor the most recent developments carefully and keep you updated.

Authors: Cristina Vila and Mireia Prat

This post is also available in: Español



82 artículos


30 artículos