Invisible Cartels in Digital Markets: Algorithmic Pricing and the Limits of Competition Law

by Tripti Gaur (Damodaram Sanjivayya National Law University, Vishakhapatnam)

Keywords: Algorithmic collusion; Competition Act 2002; Competition Commission of India (CCI); Cartels; Tacit coordination; AI pricing.

Abstract

The prompt adoption of autonomous pricing algorithms has shaken foundational assumptions of cartel enforcement. Such systems can generate prices that remain persistently higher than what a competitive market can produce, while avoiding any express human agreement, communication or any intention to collude features which would traditionally fall under antitrust liability. The question of law arises, whether algorithmic parallelism and self-learning pricing systems can amount to ‘Cartels’ under Indian Competition law, especially section 3(3) of The Competition Act, 2002 (The Act), this section presumes some coordinated outcomes to be anti-competitive. Through comparative analysis of enforcement approaches adopted by the Competition Commission of India (CCI), the U.S Federal Trade Commission and The European Union, this blog will highlight emerging evidentiary challenges, doctrinal gaps and regulatory responses to AI-enabled coordination, with the argument that machine driven collusion leads to a growing disconnect between traditional intent-focused framework of competition laws and outcome-driven realities of digital markets. The resulting disconnect highlights the importance of updating the existing enforcement standards, for it to remain effective in evolving digital spheres.

Introduction

Digital markets are largely dependent on dynamic pricing algorithms to improve efficiency, respond to demand fluctuations and optimize revenues. While these tools aid innovation they also carry unintended risk, i.e. the facilitation of tacit collusion, which means, unlike traditional cartels which are formed through explicit agreement or meeting of minds, these algorithmic systems independently learn to mirror competitors pricing strategies, thus, stabilising prices above competitive levels without any human coordination, simply put,  this enables competitors to quietly coordinate through autonomous algorithms that learn to keep prices high without any human intent or any direct communication. This phenomenon challenges the foundational assumptions of cartel law. Under Section 3 of The Act, Price-fixing agreements are illegal; the statutory framework assumes the presence of human intent and communication, however, autonomous algorithms unsettle this assumption. The self-learning reinforcement model can conclude at a collusive equilibrium despite being explicitly programmed to act independently. This blurs the traditional boundary between legal conduct and collusion; if no human ‘meeting of minds’ occurred, can the outcome still be treated as an antitrust violation? The CCI’s 2025 AI Market Study has recognised such risks and cautions that algorithmic pricing tools may mimic cartel-like outcomes without human interference.

Moreover, this debate is not unique to India; across various jurisdictions, competition authorities have struggled with algorithm-enabled anti-competitive conduct under traditional cartel laws. This highlights the need for firmer doctrinal frameworks for modern pricing technologies. Focusing on Competition (Amendment) Act, 2023, this discussion evaluates how Indian law responds to algorithmic collusion and analyses Indian approach alongside developments in the EU, US and U.K, emphasising the need to align traditional cartel principles with emerging algorithm-driven market practices.

What does it mean for algorithms to “collude”?

Algorithmic collusion is a situation which occurs when competing firms use pricing algorithms powered by A.I, without any human agreement, which behaves like a cartel. Such a phenomenon is achieved by constantly monitoring rivals’ prices and adapting in real time; these systems can gradually settle on prices that stay higher than competitive levels. For instance, two e-commerce sellers may use similar pricing software to continuously match their price hikes with each other, settling on a higher equilibrium without ever communicating, such collusion is more effective and invisible than humans, leaving no documentary trail.

This raises a central legal issue, i.e, traditional competition law targets explicit agreements to prove collusion but algorithmic coordination blurs the line between lawful parallel pricing and unlawful implicit coordination because there is no documentary trail, intent or direct contact which complicates enforcement under current legal frameworks that require proof of agreement.

Such coordination may also arise through shared digital infrastructure, In Eturas case by the European Court of Justice, it was held that participants in a common software platform may be found to have engaged in a concerted practice if they are aware of the anticompetitive message transmitted through the platform and fail to publicly distance themselves from it, but the mere existence of the platform and technical restrictions is not sufficient in itself to establish such practice. Similar dynamics were observed in US v. Topkins and in the UK in Poster & Frames, where algorithms were used to implement and monitor collusion, The CCI’s recent studies have likewise warned that ‘value-driven’ AI systems can quietly replace explicit collusion, leaving traditional detection tools ineffective.

Moreover, algorithmic collusion can emerge through multiple mechanisms from reactive pricing, where algorithms continuously observe and match each other's prices to advanced self-learning systems that optimize long-term profits, complicating detection and raising questions of firm liability for the outcomes they generate. Automated coordination adjusts prices faster and more transparently than humans, which raises concerns about fair competition and consumer impact in digital markets.

Competition Law in India: Cartels and Tacit Collusion

The Act prohibits cartels involving price-fixing, output limitation or market allocation, resulting in such conduct being illegal and anti-competitive under Section 3(3). However, liability still lies on the existence of an ‘agreement’ which Indian law has traditionally treated as different from firms simply acting in parallel.

Algorithmic pricing complicates this distinction as the firms may independently deploy software that generates collusion without any communication with invisible trails. In Amir Agarwal v. ANI Technologies, the CCI rejected allegations that Ola and Uber’s surge-pricing algorithm amounts to cartelisation, holding that use of common algorithms, by itself, does not establish concerted action. This decision of CCI reflects their reluctance to infer agreement solely from algorithmic outcomes.

At the same time, enforcement of Indian laws is slowly evolving. In Matrimony.com v. Google, the CCI observed that search algorithms could be used to manipulate ranking and result in abuse of dominance, showing their readiness to examine closely at code-driven conduct. The CCI 2025 Digital Market study also highlights that even basic algorithms can settle prices higher than competitive levels. Additionally, the Competition (Amendment) Act, 2023, strengthens digital market oversight, noting that traditional evidentiary standards may no longer work in Algorithm -led markets.

International Perspectives

1. United States: Section 1 of the Sherman Act highlights that liability depends on proving the existence of an agreement. Parallel pricing by algorithms is considered lawful despite no human coordination. In United States v. Topkins, sellers first agreed not to undercut one another and then used AI driven tools to enforce that deal and set up pricing, it was decided that the violation lay in the human agreement and not the software. Similarly, in the case of Meyer v. Kalanick Uber's surge-pricing algorithm created uniform price signals, but the claim failed as there was lack of any evidence of conspiracy. These cases reflect that US laws remain intent-centric.

   
   

2. European Union: Article 101 of the Functioning of the European Union (TFEU) covers both implicit coordination and explicit agreements. In Eturas, travel agencies using common software were held liable, despite the lack of direct communication; the court put importance on their understanding of imposed discount caps. The Commission has also treated algorithmic price monitoring as resale price maintenance, as observed in the 2018 cases against Asus, Philips, and Pioneer. Hence, reflecting that the EU takes a more outcome-focused approach.

   
   

3. United Kingdom: The Competition and Markets Authority (CMA) has conducted studies which show that pricing algorithms can facilitate coordination in hub-and-spoke collusion arrangements without any human coordination or communication, but is fair there has been no case where it has been held that pure algorithmic tacit collusion are illegal.

Enforcement Challenges and the Way Forward

Algorithmic collusion poses serious enforcement challenges for competition authorities because it falls outside the traditional antitrust framework designed to detect explicit human agreements, leading to issues such as: -  

1. Lack of Detection and Evidentiary Trail - Algorithmic collusion does not leave evidentiary trail such as meeting of minds, a written agreement like traditional cartels do. Algorithmic collusion can happen automatically through machine learning systems. Without an evidentiary trail, authorities see only similar pricing, which makes detection difficult.

   
   

2. Inadequate Technical Standards - Many competitor agencies lack the tools and expertise to analyse complex AI systems. Algorithms often function as ‘black boxes’, making their pricing decisions difficult to comprehend, which hinders effective investigation and monitoring.

   
   

3. Difficulty in Ascertaining Intent - Competition law mandates proof of intent or agreement. However, when decisions are automated, proving deliberate collusion becomes difficult, creating uncertainty about liability.

   
   

4. Rapid Market Dynamics and Real Time Pricing - Digital markets move quickly, with algorithms adjusting prices in just seconds, anti-competitive outcomes may occur and vanish before regulators can react, making traditional investigations too slow.

   
   

5. Attribution and Accountability Issues - It is often unclear who should be held responsible: the firm, managers or developers. Blurred responsibility makes it difficult to clearly assign blame and enforce sanctions effectively.

To effectively address the unique enforcement challenges of algorithmic collusion, competition authorities must adopt practical tools and legal standards designed for digital markets rather than solely relying on traditional intent-based frameworks, including: -

1. Mandatory Algorithm Auditing and Transparency - Regulators should mandate structured documentation and third-party audits of pricing algorithms to ensure transparency and accountability. Such audits can detect price collusion where traditional evidence may be lacking.

   
   

2. Foreseeability-Based liability - Competition law should implement liability where it is foreseeable that algorithms enable coordinated pricing without direct human involvement. Regulators may step in when companies show similar pricing patterns and shift the burden to firms to prove their competitive conduct.

   
   

3. Technical Expertise without Enforcement agencies - Regulators must invest in exclusive technical units with expertise in data science, AI and economic analysis to interpret algorithmic behaviour, detect harmful patterns through simulations.

   
   

4. Industry-led standards and Self-Regulation - Apart from regulatory mandates, industry groups and associations should develop and promote best practice standards for algorithmic design, documentation and third-party assessment. Promoting the adoption of internal compliance and self-audit protocols, such as documenting how AI tools reach pricing decisions and identify potential collusive patterns which can help embed healthy competition practices.

   
   

These measures shift enforcement from a case-by-case approach to a forward-looking, structured regime that can detect, prevent and address algorithmic collusion in digital markets.

Conclusion

Algorithmic collusion presents traditional laws with unprecedented challenges, as automated pricing tools take coordination to unforeseen levels. With significant attention set on advanced collusion and coordination, Indian laws, like many anti-trust regimes, have yet to move away from requiring an ‘agreement’ or ‘concerted practice’ for a cartel to exist. This gap between law and tech reality is widening. Algorithms 'behave' differently from humans, and regulators shall prepare for digital ‘meetings of mind’ that occur in code rather than boardrooms.

The future enforcement does not lie in rewriting competition law entirely, but in implementing a more effect-based analysis in competition law. As pricing algorithms rapidly grow more sophisticated, the law must ensure that collusion, whether human or machine-generated, does not escape scrutiny.