CCI’s AI shift to rekindle antitrust framework

CCI’s AI shift to rekindle antitrust framework

It signals where future competition work will emerge, as systems become central to market operations

The recently released report of the Competition Commission of India’s (CCIs) artificial intelligence (AI) market study has added a new area that could reshape the demand for competition advisory work. It introduces algorithmic transparency, self-audit, and competition-by-design, signifying that companies must proactively ensure their AI systems do not distort competition.

Dhanendra Kumar, CCI’s founding chairperson, stated, “The AI market study marks a forward-looking shift. It recognises that market conduct is increasingly shaped by algorithms, and India’s competition framework has to evolve with that reality.”

This means that companies that previously depended on case-based mandates will now have to build capabilities in ongoing compliance, algorithmic risk assessment, and governance advisory – the work, which blends legal, economic, and technical expertise.

While this will not translate into immediate new business, it signals where future competition work will emerge as AI systems become central to market operations.

After years of lull, the Commission’s AI study offers the antitrust bar a new commercial frontier, prompting firms to build cross-disciplinary teams fluent in code and competition. The CCI’s enforcement had slowed in the past few years, pushing firms to rely on merger control for a continuous revenue stream.

As large penalties that once drove antitrust revenue and urgency faded, it left enforcement lawyers with few legacy cases and low-stakes filings. However, cartel investigations picked up, bringing some long-dormant cases back.

The CCI study identifies where competition breaks down in AI markets. The survey revealed that 68 percent of startups cited data access as the primary barrier to entry, with 61 percent pointing to the costly cloud services. While self-learning algorithms may facilitate collusion without human intent, dominant platforms could self-prefer their own AI tools or tie them to cloud contracts. Similarly, dynamic pricing algorithms, used by 52 percent of respondents, showed discrimination.

The CCI has initiated a six-pillar self-audit framework that covers governance, algorithm design safeguards, testing protocols, monitoring systems, transparency measures, and documentation.

Thus, companies deploying AI for pricing, recommendations, or supply chain optimization must audit whether their systems harm competition. This shift in philosophy from reactive investigations to preventive monitoring, creates immediate demand for compliance consulting.

Nisha Kaur Uberoi, head of antitrust at JSA Advocates & Solicitors, remarked, "The CCI's call for AI self-audits and algorithmic transparency has made technical fluency a core compliance skill and will lead to retooling across the antitrust bar.

She added, for instance, e-commerce companies integrating dynamic pricing should now look at competition compliance reviews by lawyers who understand algorithms before model deployment.

However, practitioners cautioned against expecting quick windfalls.

Rahul Rai, co-founder of antitrust boutique Axiom5, felt that existing firms with big tech clients would capture initial enforcement work. He reflected, "No new practice in the immediate short run. It will take the trigger of the CCI opening an investigation in the AI and antitrust intersection. In the mid to long run, midsize and smaller companies will start investing in compliance.”

Meanwhile, large multinationals, especially in the tech sector, already face binding obligations under the European Union’s AI Act and the Digital Markets Act.

Rai added, “EU laws are already in place, and large tech companies are figuring out ways to comply. For India, they will look at the delta between the CCI guidance and the EU rules. A lot of this will be handled by their foreign lawyers who have been in the trenches globally with these companies across jurisdictions.”

The so-called ‘Brussels Effect,’ where Europe’s regulatory capacity sets de facto global standards, means multinational companies engineer compliance for the EU first and then make marginal adjustments for other jurisdictions. India represents future growth potential, not immediate revenue.

India’s Digital Personal Data Protection Act, 2023, imposes governance requirements on data collection, which defines a competitive advantage in AI markets. As algorithmic price discrimination raises competition and data protection concerns, the Commission plans a memorandum of understanding (MoU) with other regulators to coordinate investigations.

The CCI chairperson highlighted, "Traditionally, antitrust lawyers focused on market power, consumer harm, and anti-competitive agreements. However, in the digital economy, these issues are increasingly intertwined with how companies collect, process, and use personal data. Lawyers will need to understand algorithmic decision-making, data governance frameworks, and the techno-legal architecture of consent and anonymization."

With some companies restructuring accordingly, Ravisekhar Nair, partner and antitrust head at Economic Laws Practice, remarked, "We are using conversations with clients concerned about the DPDP Act to also talk about AI. Over the last 12 months, we added additional layers to our pure-play antitrust offering to cover the convergence of antitrust and data privacy and digital regulation. We have brought in the right resources. Combining data privacy digital regulation with pure play antitrust offering expands mandates and addresses client needs.”

Commenting on the CCI study, Anshuman Sakle, partner at Khaitan & Co, said that it emphasized the capabilities "beyond traditional economic and legal analysis. How lawyers acquire technical fluency in machine learning and algorithmic design varies. This will require us to work with forensic companies that have tech capabilities. In the past, a few firms had investigative tools available internally, and that practice might grow."

Meanwhile, Economic Laws Practice screens candidates for AI proficiency. As Nair stated, "I am very particular about talking to recruits about their comfort with AI. I insist that new hires be comfortable using AI tools to help with research and drafting."

Similarly, boutiques emphasized client-led training. The client-as-educator model offers cost-effective knowledge transfer for understanding specific products.

As Axiom5’s Rai explained, "We invite experts from within our clients to come and speak. We invited people from the industry to explain the AI stack. We conduct sessions where clients teach us their side of things."

JSA Advocates’ Uberoi identified required competencies - competition economics, studying AI's impact on prices, machine learning interpretability tools, and data audit frameworks. He stressed, "Law firms and data scientist teams will need to start working together, with lawyers assessing legal risks while engineers test how algorithms behave."

On the other hand, CCI’s Kumar suggested tracking US and European enforcement.

He stated, "Regulators in Europe and the US are moving towards active intervention along the vertical AI stack. Legal practitioners should anticipate that similar cross-layer enforcement trends will emerge in India."

Investigations into vertical integration, whether dominant firms leverage control of cloud infrastructure or foundational models to advantage their own applications, offer templates for Indian cases. For India, where competitive strength lies in applications rather than foundational models, foreclosure risks are salient.

Rai observed, "India's tech advantage lies in applications built atop LLMs. Assume a large cloud service provider is also present in the app layer, with all applications distributed through clouds. Then, if they favor their own applications to the exclusion of competing apps, it could potentially result in a foreclosure of competition."

While noting that remedies should extend beyond fines, Modhulika Bose, partner at Chandhiok & Mahajan, stated, "The CCI has wide powers to impose behavioral as well as structural remedies, including mandating divestments or creating information firewalls. Thus, practitioners should anticipate mandated non-discriminatory access terms or interoperability requirements, she added.

The conversation shifted to algorithmic collusion, where self-learning pricing algorithms independently coordinate without human intent, and poses unique evidentiary difficulties. Traditional cartel cases rely on documentary proof of meetings or communications. When anti-competitive outcomes emerge from opaque machine-learning systems, the smoking guns disappear.

Bose held, "Algorithmic scrutiny is not new. Indian antitrust lawyers, having dealt with cases in ridesharing, airline pricing, and e-commerce, are well-equipped. But the push towards self-auditing means even those yet to face CCI scrutiny should proactively seek legal assistance at the product development stage."

In aviation cases, human oversight of final pricing established commercial independence. As algorithms become autonomous through deep reinforcement learning, that defense erodes.

Rai argued, "It's an interesting area, but at the end of the day, it's simply a human being replaced by a tech tool to facilitate collusion. The knowledge of deployment of a tool that facilitates collusion will be easier than actual meetings between people."

While proving autonomous collusion remains contentious, Khaitan & Co’s Sakle said, "Even if we assume that everyone is using the same LLM or application, what AI does in the period of learning, the info and parameters shared are different. Under these circumstances, it will be difficult to conclude that there is tacit collusion. The Commission will find its way over a period to figure out what level of evidence will be sufficient, and the judiciary will also weigh in."

Meanwhile, Bose cautioned that opacity will not constitute a defense. "The airlines case offers guidance where human oversight dispelled collusion concerns. But opacity in algorithmic logic won't be foolproof defence; companies should align with auditing and accountability frameworks when developing tools or deploying third-party software."

The CCI study stressed that it could not single-handedly revive a slowing antitrust practice. Enforcement must materialise, penalties must sting, and clients must perceive genuine regulatory risk. The business model challenges: commoditised merger work, negligible enforcement in traditional sectors, conflict walls limiting big tech mandates, will not disappear overnight.

Still, the study offered a blueprint. It expanded practice beyond reactive litigation into proactive compliance, demanding technical expertise that may de-commoditise work, integrating antitrust with adjacent digital regulation.

The CCI chairperson warned, "AI markets will not be left to self-regulate. Lawyers must help design compliance-by-design frameworks ensuring openness, interoperability, and equitable data access to mitigate future enforcement risk."

Uberoi insisted, “The opportunity lies in positioning before enforcement actions materialize. Early consulting firms that build AI compliance toolkits, software to test and report algorithm fairness, will gain a strong competitive advantage and ensure robust defense in case of any future investigation.”

Meanwhile, whether the regulator follows its guidance with substantive enforcement remains uncertain.

However, after years of dormancy, the Commission has articulated a vision for competition law in the age of AI. The antitrust bar vision offers a potential path forward if practitioners can acquire the technical skills.