CCI's AI Market Study: What India's competition bar needs to prepare for
India's competition regulator has put AI on notice. The Competition Commission of India (CCI) released its Market Study on Artificial Intelligence and Competition, highlighting that algorithms can collude without human intervention and may be used for exploitative conduct. It also flags the dominance of Alphabet, Amazon, Meta, Microsoft, OpenAI, and Nvidia across data, compute, and foundational models-while startups face steep entry barriers.
Competition law expert Vivek Agarwal, who leads the antitrust boutique CompLaw and has helped draft CCI regulations and amendments to the Competition Act, weighed in on what this study means for enforcement and corporate compliance. Here's the distilled brief for legal teams.
A cautious, enforcement-ready stance
The study takes a measured approach: encourage self-regulation now, avoid premature hard rules, and preserve innovation. This mirrors the CCI's 2020 e-commerce market study, which later informed enforcement. Expect the report to guide how the CCI engages with government on the in-progress digital competition law and to frame AI-specific scrutiny without freezing useful adoption.
Where AI stresses current antitrust tools
- Algorithmic collusion without human coordination challenges traditional Section 3 analysis.
- Self-learning and generative models reduce traceability and accountability.
- Startups struggle to access quality data, cloud services, talent, and funding-while large incumbents control essential infrastructure, proprietary models, and datasets.
- The regulator must build deeper technical capacity to supervise and investigate AI-driven conduct.
Self-regulation: the immediate ask
The CCI expects firms to embed competition checks inside AI systems. If models operate as black boxes, ex-post enforcement may be too slow or uncertain. Many enterprises already deploy AI for demand forecasting, inventory planning, and personalization-useful, but not risk-free.
Two risk clusters stand out: AI pricing that drifts into tacit coordination, and unilateral conduct by dominant firms-self-preferencing, predatory pricing, and discriminatory pricing-amplified by automated decision systems.
What actions to expect from the CCI
- Advocacy and capacity-building: Focused workshops to raise compliance standards and encourage pro-competition innovation.
- Inter-regulatory MoUs: Coordination with sectoral regulators to reduce gaps and overlaps.
- Bundling scrutiny: Closer look at integrations of proprietary AI tools (e.g., Copilot, Gemini) into established products where such bundling may entrench dominance or foreclose rivals.
- Deal review: Tighter lens on strategic acquisitions and partnerships in AI that could cement control over data, compute, or distribution.
Practical steps for in-house legal and counsel
- Map AI use-cases and risks: Keep a live inventory of models and tools. Document objectives, inputs, constraints, and human oversight for each system.
- Guardrails for pricing tools: Prohibit shared sensitive data, common optimization parameters, or vendor settings that could facilitate parallelism across competitors. Keep audit logs and changelogs.
- Vendor and API diligence: Assess model providers for data sources, training methods, and tuning practices. Build in competition-compliance clauses, audit rights, and kill-switch obligations.
- Data controls: Ring-fence competitively sensitive information. Restrict cross-use between business units and products. Monitor feature updates that change data flows.
- Dominance checks before launch: For platform integrations, run bundling/tying and self-preferencing tests. Evaluate switching costs, default settings, interoperability, and access for independent rivals.
- Fair-pricing reviews: For personalized or dynamic pricing, test for unjustified discrimination and exclusionary outcomes. Keep clear rationale and evidence of consumer benefit.
- Merger control readiness: Screen AI-related investments and partnerships early for possible notification triggers. Track minority stakes, data-sharing terms, and exclusive arrangements.
- Incident response: Define thresholds for pausing or rolling back algorithmic deployments. Escalation paths should include legal, product, and data science.
- Capability building: Train legal and product teams on algorithmic accountability and antitrust risks in AI deployment. Consider role-based AI upskilling for legal teams via resources like AI courses by job.
Why this matters now
The study signals heightened attention on AI uses that tip markets, foreclose entry, or blur accountability. It also invites firms to show they can police themselves. Those who establish credible governance-model documentation, data discipline, and reviewable decision logic-will be better placed if the CCI moves from study to enforcement.
Further reading
Source perspective: Insights reflect the views of competition law expert Vivek Agarwal, head of CompLaw, who has contributed to CCI regulations and key antitrust cases across cartels, abuse of dominance, vertical restraints, and combinations.
Your membership also unlocks:
