AI and EU Competition Law: The Next Frontier of Antitrust Enforcement

Introduction: When Algorithms Compete — or Collude

Artificial intelligence and machine learning are reshaping how businesses set prices, allocate resources, and compete. But they also raise a new and complex question: can algorithms collude?

As digital markets evolve, the European Commission’s Directorate-General for Competition (DG COMP) faces challenges unimaginable when EU competition law was codified decades ago. Traditional antitrust tools—built for human decision-makers—are now being tested against autonomous, data-driven systems capable of coordinating behavior without explicit human intent.

The intersection of AI and competition law represents one of the EU’s most intellectually demanding and legally significant frontiers.

---

Article 101 TFEU and Algorithmic Collusion

Article 101 of the Treaty on the Functioning of the European Union (TFEU) prohibits agreements and concerted practices that restrict competition. The challenge is determining whether AI-driven market behavior can amount to “coordination” in the absence of human communication.

Types of Algorithmic Collusion

Legal scholars and regulators identify several emerging models:

1. Messenger Collusion – humans design algorithms to implement an illegal cartel (e.g., coordinated pricing).

2. Hub-and-Spoke Systems – a shared pricing algorithm provider enables indirect collusion among competitors.

3. Predictive Collusion – independent algorithms learn to align prices through data observation, without explicit communication.

The third scenario—“tacit algorithmic collusion”—is particularly troubling. If machine learning leads to self-reinforcing parallel conduct, how can regulators prove intent, or even identify a “meeting of minds”?

DG COMP has acknowledged this problem in multiple policy papers, noting that traditional evidentiary tests may be insufficient in algorithmic markets.

---

Article 102 TFEU: Abuse of Dominance in Digital Markets

AI also affects how dominant firms may abuse market power. Under Article 102 TFEU, practices such as self-preferencing, exclusionary data use, and algorithmic manipulation can constitute abuses.

Examples include:

Self-Preferencing: A dominant platform’s algorithm favors its own products (e.g., Amazon Marketplace ranking AmazonBasics higher).

Exclusion via Data Control: Large AI firms restrict data access to disadvantage rivals.

Dynamic Pricing Exploitation: Personalized pricing algorithms extract maximum consumer surplus, raising fairness and transparency concerns.

The Digital Markets Act (DMA) now codifies several prohibitions—turning long-debated competition concerns into ex-ante obligations for “gatekeepers.” Yet DG COMP continues to play a role in case-by-case enforcement when abuses extend beyond DMA thresholds.

---

Merger Control and AI Market Concentration

AI development has accelerated strategic acquisitions by tech giants—often of small, innovative startups in data or analytics. Many of these deals fall below traditional turnover thresholds, yet have profound long-term competitive effects.

In response, the Commission and national authorities have adopted:

Article 22 EUMR referrals, allowing Member States to refer small but significant acquisitions (e.g., the Illumina–Graill case),

Closer scrutiny of “killer acquisitions” where incumbents absorb potential rivals before they grow,

Focus on data-driven mergers, where combining datasets may entrench dominance even without price effects.

AI models thrive on data; thus, mergers that consolidate massive data pools can create barriers to entry and innovation choke points. Competition authorities are now developing new analytical frameworks to assess these non-price dimensions.

---

Data, Access, and Interoperability: The New Competitive Currency

In the AI economy, data is the new essential facility. Companies that control vast data ecosystems can limit rivals’ ability to compete.

EU law is evolving to address this through:

The Data Act (2023), granting users and competitors rights to access industrial data,

The Digital Markets Act, requiring gatekeepers to ensure data portability and API interoperability,

Competition law doctrines of refusal to supply and essential facilities, now interpreted for data-based markets.

Together, these instruments form an emerging “data competition regime”, where control over information—not just price or output—defines dominance.

---

AI Bias and Antitrust: When Algorithms Shape Consumer Choice

Competition law traditionally assumes rational, informed consumers. But AI-driven recommendation systems can subtly shape preferences, steering users toward certain products or services. This “algorithmic nudging” blurs the line between competition and consumer manipulation.

Regulators are beginning to explore whether such systems—especially those leveraging dominance and asymmetric data—could amount to:

Exploitative abuses under Article 102 (manipulative design, attention capture),

Consumer harm justifying DMA enforcement,

Cross-sector coordination with data protection (GDPR) and consumer law (Unfair Practices Directive).

This marks a shift toward a behavioral antitrust model, where competition law and digital ethics increasingly overlap.

---

The Commission’s Response: Guidance, Studies, and Enforcement

DG COMP and national competition authorities are preparing for the AI era through:

The European Competition Network (ECN) working group on digital algorithms,

Joint studies with the OECD and CMA (UK) on algorithmic collusion,

Integration of AI expertise and data analytics into enforcement teams,

New Guidelines on Horizontal Cooperation Agreements (2023) addressing algorithmic price coordination.

Enforcement remains cautious—regulators are reluctant to stifle innovation—but the direction is unmistakable: algorithmic transparency and accountability are now competition concerns.

---

Conclusion: The Future of Antitrust in the Age of AI

As artificial intelligence transforms markets, the EU faces a delicate task—balancing innovation with fair competition. Traditional antitrust principles remain sound, but their application must evolve to meet algorithmic realities.

Future competition law will likely require:

Explainability obligations for AI systems influencing market conduct,

Algorithmic audits as part of merger and compliance reviews,

Cross-regulation coordination between competition, data, and AI authorities.

In the EU’s model of digital governance, AI and competition law are converging—creating a framework where innovation thrives within the rule of law, and algorithms serve markets rather than manipulate them.