India is preparing for a mega global technology event in the month of February 2026, “The India AI Impact Summit 2026”. The summit gives an opportunity to discuss implementation of a collaborative framework for regulation of AI in India and discussing the role played by AI in economic and social transformation. Against the heightened national focus on the governance of AI and policy coordination, regulatory institutions in India such as the Competition Commission of India (“CCI”) through its market study have outlined their position on risks and opportunities presented by AI. The CCI has released its report titled ‘Market Study on Artificial Intelligence and Competition’ (“Report”), highlighting the potential competition issues caused by Artificial Intelligence (“AI”). The Report marks India’s first attempt at examining the impact of AI on market dynamics and outlines the regulatory actions required to address these challenges.
The short piece examines the approach adopted by the CCI in navigating competition law concerns emerging from India’s rapidly evolving AI landscape. The authors highlight the regulatory approach adopted by different regulators, while contrasting the approach adopted by the CCI. Then, the authors scrutinise the self-audited framework proposed by the CCI. Lastly, the authors highlight the practical challenges that the CCI could face in the future by adopting the self-audit framework and provide a few suggestions to navigate these challenges.
The European Union (“EU”) has adopted a proactive approach to AI regulation by implementing the Artificial Intelligence Act, Digital Markets Act (“DMA”), Data Act and Digital Services Act. The DMA regulates the conduct of core platform services (“CPS”) offered by gatekeepers. The DMA could regulate AI in two ways: (i) the AI developer may provide CPS and meet the requirements of a gatekeeper under Section 3 of the DMA, and (ii) AI-powered systems may be integrated into CPS, thereby obliging them to abide by the obligations imposed under the DMA.
On 23 January 2025, President Trump signed a new executive order on Removing Barriers to American Leadership in Artificial Intelligence. This policy is focused on the development of ‘unbiased and agenda-free’ AI systems and investments in AI infrastructure. Although the United States does not have a federal antitrust law regulating digital markets, the Federal Trade Commission (“FTC”) has been proactive in investigating antitrust concerns in the digital space. In January 2024, the FTC launched an investigation into generative AI investments and partnerships. The FTC and Department of Justice (“DOJ”) are investigating the Big Tech companies. The DOJ investigated Nvidia and its dominant position in supplying semiconductor chips for AI models. Likewise, the FTC is set to investigate Microsoft and OpenAI for having an unfair advantage in the AI space, especially for LLMs.
While major jurisdictions like the EU have adopted a proactive approach in regulating AI models and systems by placing strict obligations on developers and marketers of ‘high-risk’ AI systems, other Asian countries, such as Japan, Singapore, and the United States, have adopted a ‘cautious’ regulatory stance. These countries have implemented a voluntary framework, such as guidelines, to promote the responsible use of AI, aiming to balance responsible deployment with the promotion of innovation, competition and economic growth.
In contrast to countries such as the EU, which have adopted an ex-ante framework, the CCI seems to have adopted a ‘measured and pragmatic’ approach, consistent with other Asian countries and the United States. This pragmatic approach is focused on providing an overarching framework for AI governance that promotes responsible deployment of AI while protecting innovation and market competitiveness.
Key aspects of the report
The Report emphasises on three key aspects for protecting market competitiveness and innovation in the AI realm, (i) implementation of a self-audit framework of AI systems for competition compliance, (ii) removing barriers to entry by expanding national AI computing infrastructure and access to high-quality, non-personal datasets thereby enabling startups to compete with Big Tech companies, and (iii) enhancing the ability of CCI to develop expertise in AI technologies and setting up a think tank of experts to advice on competition concerns in digital markets and AI. The Report also recognises the importance of inter-regulatory coordination, emphasising the need for a unified approach to AI governance through the execution of inter-regulatory Memorandums of Understanding (“MoUs”).
Step in the right direction
By implementing a self-audited framework comprising continuous monitoring, control and audits of AI algorithms, system mapping and identification of high-risk factors adversely impacting competition, the CCI has taken a pivotal step towards effective AI governance. This framework provides companies with an opportunity to demonstrate that their AI systems are not likely to cause an appreciable adverse effect on competition (“AAEC”) under Section 3 of the Competition Act 2002. The documentary proof of internal safeguards, such as the establishment of a competition compliance program, identification of high-risk systems and determining the manner of interaction of AI systems with different market participants, acts as a vital safeguard for companies functioning in a rapidly changing AI ecosystem, where laws regulating AI in India are being debated.
Additionally, this self-regulation framework enables the CCI to consider the internal compliance standards and provisions that companies have implemented to stop anti-competitive behaviour such as algorithmic collusion, self-preferencing, algorithmic exploitative conduct, etc. The CCI may take into account such evidence when determining the penalty computation under its Penalty Guidelines, potentially minimizing liability and underscoring the value of good-faith compliance in a complex and changing AI environment. However, certain critical concerns regarding the practical enforceability of this self-audited framework may be considered.
Regulatory overlap
Firstly, the CCI is yet to clarify which regulator would have first jurisdiction. The question arises as to whether the CCI would have first jurisdiction in cases involving overlapping concerns such as competition, IPR, cybersecurity, privacy, etc. The Report states that “the CCI to enter into MoUs or arrangements with any statutory authority or department of the government for the purpose of discharging its duties or performing its functions under the Act”. The existence of a sector specific regulatory body such as Telecom Regulatory Authority of India (TRAI) for the telecom sector or the Controller General of Patents for patents, raises the question as to “who would have first jurisdiction to examine the case in the event the CCI and the specific regulatory body enter into an arrangement/ MoU?” In practice, the CCI has often claimed first jurisdiction over TRAI and the Controller General of Patents for cases having overlapping concerns.
In Telefonaktiebolaget LM Ericsson v. CCI, the CCI claimed first jurisdiction over the Controller of Patents for exercising its investigatory authority in relation to charging of excessive royalties, discriminatory conditions for licensing patents, and not making patents reasonably available. The CCI contended that the exercise of its powers cannot be prevented due to a mere ‘overlap’ between the Patents and Competition Act. The CCI emphasised that it would have jurisdiction owing to anti-competitive and abusive behaviour undertaken by the patentee. Furthermore, it emphasised that the Patent Act is insufficient to guide the Controller in examining instances of anti-competitive behaviour on the part of the patentee. The Delhi High Court did not fully ouster CCI’s jurisdiction but directed it to continue its investigation while not passing any final order until further hearing. Following the CCI’s appeal, the Supreme Court declined to interfere.
The Supreme Court, in another landmark judgement of Competition Commission of India v. Bharti Airtel Limited, ruled that the TRAI, being a specialised regulatory body, has first jurisdiction in matters relating to the telecom sector. Upon TRAI’s final determination and identification of anti-competitive practice, CCI would have jurisdiction to investigate and impose restrictions under the Competition Act.
These cases illustrate that, depending on the nature of the dispute, jurisdiction may alternately vest with sectoral regulators or CCI, leading to a regulatory overlap. Similar jurisdictional ambiguities are likely to emerge with the use of AI across different industries and sectors. Therefore, in order to avoid regulatory overlap and guarantee consistent enforcement, the CCI must precisely define its jurisdiction in the context of AI systems.
Determination of tipping point
Secondly, the Report does not provide guidance for determining the tipping point at which market dominance may arise. The large user base enables companies to collect valuable user data used for training AI agents thereby increasing the service quality and attracting additional other users. Consequently, large AI platforms such as OpenAI, Google, Microsoft and Perplexity possess the ability to compete successfully in the AI space. Contrarily, new entrants such as startups and Micro, small and medium enterprises (MSMEs) lack the user base to train and develop competitive AI models. In addition to access to data, network effects– the increase in a platform’s value as more users join–serve as a barrier to entry for companies.
AI platforms are dependent on both the number of users and their data. The increase in users leads to more diverse data, leading to better output by AI models. Network effects arise when the increase in value of a product/ service to a user draws other users, leading to an increase in market share for companies. Positive network effect leads to tipping in favour of one company, resulting in a ‘winner-takes-all’ outcome and creating information asymmetry. The CCI must clarify the threshold at which such information asymmetry leads to tipping– market dominance.
The Report also acknowledges that AI has been adopted in different sectors such as retail, healthcare, e-commerce, marketing, banking and financial services. However, the CCI has not clarified how it would determine the tipping point across different sectors. It remains to be seen whether the threshold for determining the tipping point would be sector-specific or application-specific. Relying on an application-based method for determining the threshold for the tipping point may result in false positives, as it would entail comparing applications in different sectors with similar features instead of ascertaining market dominance by a single company in a specific sector.
Way forward
- Determination of tipping point for AI systems
While determining the threshold for the tipping point in different sectors, the CCI will have to be wary of the nuanced distinction between algorithmic collusion and web scraping. Web scraping is a practice adopted to train AI models by automatically extracting publicly available data from multiple platforms. However, it may resemble algorithmic collusion when algorithms use comparable mechanisms to continuously monitor competitors’ pricing or strategic behaviour. For determining the threshold for market dominance, the CCI must make a clear distinction between acceptable data aggregation and collusive behaviour, given this overlap in technological approaches.
The CCI needs to provide adequate guidance for the determination of the tipping threshold and as to when tipping takes place in a particular sector. The dynamic nature of AI requires the CCI to take a flexible and sector-specific approach by establishing distinct thresholds that represent the competitive dynamics and degree of AI adoption in each industry. For providing adequate guidance to both current and new market participants, these thresholds may be reviewed on a monthly or quarterly basis, factoring in the competitiveness caused by the adoption of AI systems in different sectors and the rapid advancements in AI technology.
In the proposed self-audited framework, the CCI should require companies to disclose the individuals responsible for making decisions for the AI algorithm. This would enable the CCI to identify those responsible for algorithmic collusion and ensure accountability for anti-competitive practices. The CCI, through its FAQ’s or guidance notes, may provide guidance to companies by releasing a non-exhaustive, indicative list of situations highlighting permissible algorithmic behaviour not amounting to algorithmic collusion.
2. Self Audit of AI system
Annexure I of the Report outlines a self-audit framework for AI systems based on six pillars: governance, algorithm design, testing, monitoring, transparency, and compliance integration. However, the Report does not clarify what internal compliance entails in practice. The guidance neither specifies what constitutes adequate documentation nor specifies the role and obligations of those deploying AI systems. Thus, companies may place reliance on their own subjective evaluations rather than quantitative criteria, leaving a regulatory gap where compliance is determined by interpretation rather than strict adherence to the rules. The CCI could take inspiration from Section 119 of the Companies Act 2013, which mandates every company to maintain minutes of both the general and board meetings. The book of minutes, an internal record maintained by the company, helps ensure accountability and legal compliance. Similar to the book of minutes, the CCI may make it mandatory for every company adopting AI systems to record the risks caused by AI systems, individuals responsible for making decisions for the AI algorithm and other disclosure obligations required to assess anti-competitive behaviour. This internal book of minutes would ensure accountability by companies operating in a rapidly growing and underregulated AI environment.
(Views are personal)
