The use of artificial intelligence has radically transformed the digital market by altering how companies price, distribute resources, and engage with competitors. Unlike the classical market, where consumers make human choices, AI-based products rely on self-learning algorithms that continuously process vast amounts of data, anticipate consumer behaviour, and adjust strategies in real time. Digitalisation can offer several pro-competitive advantages. Market contestability and fair practices promote innovation and the development of new products and services. However, a strong governance framework is essential to support a smooth expansion of the digital ecosystem and to mitigate potential anti-competitive risks.
In March 2024, the Draft Digital Competition Bill, 2024 (the “Bill”) was published by the Ministry of Corporate Affairs for public consultation. The fundamental structure of the Bill is based on the determination and regulation of the Systemically Significant Digital Enterprises (SSDEs). SSDEs are digital gatekeepers with strong market power due to their data-based influence, network effects, and scale. The proposal uses a twin-test to categorise entities as SSDEs: (a) quantitative threshold: financial strength and user base, and (b) qualitative criteria: market influence, not just in raw numbers.
These developments, however, reveal a more underlying regulatory dilemma. This then poses an essential question for contemporary competition law: Can regulators effectively limit anti-competitive harms that do not arise from cartel meetings but from self-directed algorithms? This article discusses whether current and future legal systems are in a position to respond to this challenge or whether AI has surpassed the assumptions on which competition regulation usually relies.
How AI Facilitates Anti-Competitive Behaviour
The use of artificial intelligence has not only changed the nature of competition between firms but also the nature of anti-competitive behaviour in digital markets. Conventional violations rely either on the explicit agreement or intentional omission. However, AI-based harms are usually the result of autonomous optimisation, which does not always ensure easy detection and enforcement of relevant competition law frameworks.
A. Algorithmic Collusion
In algorithmic collusion, self-learning algorithms in pricing independently reach supra-competitive prices through repeated market feedback. These systems can stabilise high-priced equilibria that resemble cartel behaviour, without any human communication or explicit agreement. This negates the premise of the competition law, which is based on intent and agreement. This requires transitioning from an effects-based enforcement approach.
B. AI-based Self-Preferencing
The incorporation of self-preferencing within AI-driven ranking and recommendation systems makes the practice opaque and hard to identify. As algorithms increasingly control visibility and consumer choice on digital platforms, giving preferential treatment to a platform’s own offerings can subtly and extensively distort competition. Notably, platforms may seem to comply with non-discrimination rules while algorithmic design choices continue to produce systematically advantageous outcomes. This reveals the insufficiency of superficial regulatory compliance in tackling deeper, structural forms of algorithmic bias.
C. Predatory Personalisation and Behavioural Manipulation
AI mediates brand-customer relationships and influences customer behaviours. The dominant platforms can strategically undermine competitors in certain areas while maintaining higher prices in others, without relying on a conventional predatory pricing indicator. In the AI-based markets, damage is not even and visible but personalised and well-planned.
The combination of these practices shows a significant enforcement deficit: AI facilitates anti-competitive behaviour without any direct agreement, overt exclusion, or even transparent pricing. This is a fundamental challenge to the basic principles of conventional competition law.
India in 2026: Will the Digital Competition Law Rein in AI Gatekeepers?
The regulatory environment of digital markets in India is in limbo. Since the ex-post competition framework of the current Competition Act, 2002, is unsuitable to address the pace and the structural transformation of digital ecosystems. The Bill, however, is not deprived of limitations and ambiguities.
A. Self-Preferencing v. AI Opacity
Although the Bill in Section 12 bans SSDEs from using non-public data to compete in their main line of digital offerings, it does not prohibit its use with users’ approval. This creates a legal gap in which platforms can avoid restrictions by leveraging consent mechanisms and proceed to implement favourable algorithmic results.
B. Dynamic Markets and Static Thresholds
The two quantitative and qualitative tests may be capturing a fast-growing domestic start-up or an omission by deep-pocket foreign-based platforms with a thin India-specific financial base. The Bill does not include a strict rebuttal mechanism to oppose the SSDE designation, casting doubt on the fairness and regulatory predictability for innovators.
C. Enforcement Readiness
The policy presumes that the Competition Commission of India (CCI) will be prepared to regulate technically, such as through algorithmic analysis, data audits, and live surveillance, but its enforcement structures remain underdeveloped. Improving CCI’s technical and computational capabilities is one of the conditions for the successful implementation of the Bill. Overall, the Draft Digital Competition Bill, 2024, is a template for proactive regulation, though it still requires further improvement in areas essential to regulators when dealing with AI-facilitated gatekeepers. The CDCL’s report confirmed that the penetration of digital enterprises exhibited an uneven distribution across Indian cities.
The Way Forward: A New Regulatory Model for the Algorithmic Era
If AI has transformed markets from human-driven arenas into algorithmically mediated ecosystems, competition law must evolve accordingly. The core challenge in 2026 is not merely regulating digital firms, but regulating the algorithmic infrastructures, data pipelines, compute access, and ranking systems that now shape competition outcomes.
First, India must move towards an Algorithmic Market Power (AMP) framework. Traditional dominance analysis based on market share and pricing is insufficient in AI-driven markets where power stems from exclusive access to training data, computational resources, and proprietary models. The Digital Competition Law should explicitly recognise these inputs as sources of competitive advantage and assess dominance through an effects-based lens rather than relying solely on static thresholds.
Second, mandatory algorithmic audits should become central to enforcement. For SSDEs, compliance cannot stop at formal non-discrimination policies. Regulators must be empowered to examine ranking logic, data usage patterns, and pricing outcomes through confidential audits conducted by technical experts.
Third, India must address asymmetries in computing and interoperability. Exclusive access to cloud infrastructure and closed AI ecosystems can entrench dominance just as effectively as contractual foreclosure. Introducing principles of compute neutrality and interoperability obligations, particularly for core digital services, would lower entry barriers and preserve contestability without stifling innovation.
Conclusion
Conclusively, this article began with a central question: Can regulators control anti-competitive harms that arise not from cartel meetings, but from self-learning algorithms? The answer is yes, but only if competition law shifts its focus from human intent to algorithmic outcomes. India’s proposed Digital Competition Law marks a critical step toward ex-ante regulation of digital gatekeepers. However, effectively tackling AI-driven market power in 2026 and beyond will require competition law to go beyond traditional boundaries and explicitly regulate data and algorithms as essential sources of market dominance.