The Impact of AI on EU Competition Law
The rise of artificial intelligence (AI) raises important questions about potential violations of EU competition law. AI systems operate with varying degrees of autonomy and generate outputs—like predictions or decisions—that can influence both physical and virtual environments. This article focuses on how AI affects the enforcement of Article 102 of the Treaty on the Functioning of the European Union (TFEU), which prohibits abusive conduct by dominant companies.
Abuse of Dominance
Under EU law, a company is dominant if it can act independently of competitors, customers, and consumers to a significant extent. Holding a dominant position is not illegal by itself; what matters is whether there is abuse. Examples of abuse include actions that exclude competitors or exploit consumers unfairly.
Defining Markets
Defining the relevant market is the first step in assessing dominance and possible abuse. AI markets pose particular challenges due to the variety of technologies involved and their broad range of applications. The European Commission’s 2024 market definition notice offers limited guidance on this front.
Another difficulty is the risk of prematurely concluding dominance, especially given the geopolitical context around AI. For instance, concerns arose that U.S. companies might dominate AI markets due to their size and investments. However, the entry of new players like DeepSeek demonstrated that AI markets remain dynamic and susceptible to disruption.
Abusing Dominance Through AI
Competition authorities in Europe are especially wary of two types of abuse involving AI: self-preferencing and pricing abuses.
Self-preferencing
A dominant firm might program AI to favor its own products or services over competitors’. The European Commission fined Google €2.42 billion for allegedly favoring its own comparison-shopping service by exempting it from the usual search ranking algorithm. Similarly, Amazon faced scrutiny for using nonpublic seller data to adjust its retail strategies, harming competition.
Pricing Abuses
AI can enable dominant companies to engage in predatory pricing or price discrimination.
- Predatory pricing: Using AI to offer below-cost prices to vulnerable or price-sensitive customers in order to exclude competitors. Proving below-cost pricing remains a challenge.
- Price discrimination: AI can adjust prices dynamically based on consumer profiles or behavior. However, EU law requires strong evidence to prove exploitative abuse, including frequent and targeted discrimination without objective justification, and demonstrable harm to social welfare.
Essential Inputs
Dominant companies may abuse their position by restricting access to essential inputs like IT components and data.
Access to IT Components
Specialized chips are necessary for training AI models. A dominant chip manufacturer could engage in:
- Rebates: Offering discounts in exchange for purchase commitments.
- Refusal to supply: Denying chips to competitors in downstream markets, potentially eliminating competition.
- Discriminatory behavior: Favoring affiliated companies with better pricing or earlier access.
Access to Data
Data is critical for AI development. Smaller players may struggle to access sufficient personal data due to compliance costs, while larger companies with extensive datasets could face scrutiny for creating barriers to entry or denying access to data.
Tying and Bundling
Tying and bundling are additional ways dominant companies might restrict competition.
Tying
Tying occurs when customers must buy one product to obtain another. This can be contractual—requiring the tied purchase—or technical—where products are integrated and cannot be used separately. For example, a dominant software provider might require that its AI tool only works with its platform.
Bundling
Bundling offers discounts for purchasing a package of products. A dominant company might encourage customers to buy software bundled with AI tools at a lower price than buying separately, which can foreclose competitors.
Commitments and Enforcement
Competition authorities actively investigate potential abuses related to AI. For example, in July 2023, the European Commission opened a formal investigation into whether Microsoft tied its Teams application to its productivity software unlawfully. Microsoft responded with commitments including offering Office 365 versions without Teams and enhancing interoperability with competing products. The Commission’s assessment of these offers is ongoing.
For legal professionals interested in the evolving interface between AI and competition law, staying informed about these developments is essential. More detailed courses and resources on AI legal implications can be found at Complete AI Training.
Your membership also unlocks:
