EU Law Regarding AI and Unfair Market Practices

EU Law Regarding AI and Unfair Market Practices

The rise of artificial intelligence (AI) has introduced novel forms of potential abuses by dominant undertakings, challenging the application of Article 102 of the Treaty on the Functioning of the European Union (TFEU).

New forms of abuse via AI

Dominant firms may use AI systems to engage in self-preferencing — for example, giving preferential treatment to their own products or services — or to enact pricing abuses like algorithmic price discrimination and tacit collusion, which can harm consumers and competitors without explicit human coordination. These AI-enabled practices challenge established notions of abuse under Article 102, which prohibits dominant firms from exploiting or foreclosing markets but lacks an exhaustive list of abuses.

Challenge for enforcement

AI systems operate with varying degrees of autonomy and dynamically infer outputs from inputs, making it harder to identify intentional abusive conduct or separate lawful innovation from anti-competitive behavior. Current competition law frameworks, including Article 102, may require adaptation to address algorithm-driven exploitative behavior such as "perfect price discrimination" by AI algorithms, which remains underexplored in case law.

Challenges AI presents for market definition

Defining the relevant market is a prerequisite to assess dominance under Article 102 TFEU. The AI market is complex and often layered, involving algorithms, data inputs, and digital infrastructure, with rapid innovation and convergence of services. The European Commission’s 2024 market definition notice provides limited guidance for AI markets, reflecting the difficulty in demarcating product and geographic markets where AI firms operate.

In addition, a substantial part of AI’s computational layer, including hardware (like GPUs) and proprietary software infrastructure, remains outside direct AI regulation and can concentrate market power. For example, dominant providers of AI hardware infrastructure (e.g., NVIDIA’s CUDA platform) may exhibit monopoly features such as steep price increases, which arguably constitute abuses under Article 102 but are harder to detect and regulate due to black-box technology and lack of transparency.

The opaque nature of AI algorithms and underlying infrastructure complicates verifiability and transparency obligations, which are critical for competition authorities to investigate and prove abuse. Limited transparency constrains the ability to define markets accurately and to assess firms’ conduct within those markets.

In summary, AI introduces novel potential abusive behaviors for dominant undertakings under Article 102 TFEU, particularly around algorithmic pricing and self-preferencing, while also complicating the core competition law task of market definition due to technological opacity, multi-layered AI ecosystems, and infrastructure concentration risks. This necessitates evolving enforcement approaches and regulatory frameworks to effectively address AI-related dominance abuses in EU competition law.

References:

[1] European Commission (2021). Artificial intelligence: A European approach to excellence and trust. Retrieved from https://ec.europa.eu/info/publications/artificial-intelligence-european-approach-excellence-and-trust_en

[2] European Commission (2020). White paper on artificial intelligence - A European approach to excellence and trust. Retrieved from https://ec.europa.eu/info/publications/white-paper-artificial-intelligence-european-approach-excellence-and-trust_en

[3] European Commission (2020). European strategy for data. Retrieved from https://ec.europa.eu/info/publications/european-strategy-data_en

[4] European Commission (2019). Communication on Artificial Intelligence: A European approach to excellence and trust. Retrieved from https://ec.europa.eu/info/publications/communication-artificial-intelligence-european-approach-excellence-and-trust_en

1. In the context of AI-enabled practices challenging established notions of abuse under Article 102, there is a need to adapt current competition law frameworks, such as the prohibition against dominant firms from engaging in antitrust violations like perfect price discrimination by AI algorithms.
2. The complexity of the AI market, coupled with the opaque nature of AI algorithms and underlying infrastructure, presents challenges for competition authorities in verifying and enforcing antitrust regulations, such as Article 102, when dominant providers of AI hardware infrastructure exhibit monopoly features.

Read also: