Corporate Board Liability for Adopting Anticompetitive AI Tools:An Indian Competition Law and Corporate Governance Perspective

Abstract

In India, Artificial Intelligence (AI) has emerged as a pressing priority for businesses, especially in sectors like e-commerce, transportation, digital platforms, and financial services. AI applications are now a mainstay in industries for pricing, demand forecasting, and strategic planning. These advancements have improved operational efficiency and decision-making, but they've also introduced new hurdles in competition law enforcement. Algorithmically collusive behaviour, where the algorithms are involved in coordinated pricing and market behaviour without human intervention, is one of the major concerns. This coordination can happen organically as a result of machine learning processes, even if there is no agreement to form a traditional cartel. The development has prompted significant legal complexities about accountability and liability, especially in the context of corporate boards approving, supervising, and managing the use of AI technologies.

This paper offers a thorough exploration of the potential liability of corporate boards in terms of the anti-competitive consequences of the AI systems under Indian law. It explores the legal, judicial, regulatory and corporate governance principles, rules and frameworks to consider the extent of board responsibility. This paper outlines the array of legal and regulatory tools that can be purposefully used to address many of these concerns through governance change. It states that good compliance practices and accountability regimes are not only ;good business sense but are also increasingly becoming a legal requirement.

Keywords: Algorithmic Collusion, Competition Act, 2002, Corporate Governance, Artificial Intelligence, Board Liability, Companies Act, 2013, Competition Commission of India, Fiduciary Duties, Digital Markets.

1. INTRODUCTION

Industries across India have witnessed the rapid surge of Artificial Intelligence (AI) in just a few years, revolutionising business decision-making processes. Algorithms are now integral parts of determining prices, managing procurement, optimising logistics, recruiting employees and customer interactions, and are often acting at speed and scale greater than humans can achieve. This change has brought a whole new level of efficiency, accuracy and resource utilisation. It has, at the same time, however, generated new legal and regulatory issues that perhaps some current laws and regulations don't sufficiently cover.

A major worry is collusion in algorithms. Algorithmic collusion differs from classic cartel behaviours in that it does not necessarily require an explicit or tacit agreement among human participants, but rather can develop spontaneously by interactions between AI systems. This type of activity can be hard to identify, investigate and prosecute as it often lacks direct human contact. As a result, it can quickly become a more complicated question of who is at fault. When an algorithm exhibits anti-competitive behaviour, the next big question is: who's liable?

The present paper will concentrate specifically on the role of corporate boards, the top decision-making bodies of companies in India, in relation to anti-competitive behaviour in the context of AI systems used under their authority and control. The problem is not only competition law, but it's also corporate governance. The main issue from a competition law perspective is whether the conduct of the AI systems can be imputable to the company and its board, fulfilling the requirements of Sections 3 and 4 of the Competition Act, 2002. The corporate governance question is whether directors who approve, monitor, or are negligent in monitoring the use of AI have fulfilled their duties. The question from the corporate governance perspective is whether the directors who approved, monitored, or did not adequately monitor the use of AI have fulfilled their duties under the Companies Act, 2013.

The paper is organised as follows. Section 2 examines how Algorithmic Collusion in India. Section 3 discusses the competition law framework. Section 4 examines liability related to AI-driven behaviour. Section 5 considers directors' duties and responsibilities under the Companies Act, 2013. Section 6 specifically addresses the question of board liability. Sections 7 and 8 discuss risk management, third-party AI systems, and emerging regulatory challenges. Section 9 provides policy recommendations, and Section 10 concludes the study.

2. ALGORITHMIC COLLUSION IN THE INDIAN CONTEXT

Algorithmic collusion occurs when market participants collaborate on their market behaviour through the use of algorithms without necessitating any communication or agreement among the competing firms. While the application of algorithmic pricing has grown greatly since the advent of internet-based markets in the 1990s, economic theory had been around long before that regarding the potential for such coordination. As Artificial Intelligence (AI) becomes more prevalent, these worries have gained even more significance.

Algorithmic pricing is gaining traction in many industries in India, such as e-commerce, ride sharing, food delivery, financial services, and hotel aggregators. These businesses use real-time data analytics to dynamically adjust prices in response to fluctuations in demand, market

conditions, competitor pricing strategies and historical transaction information. Such systems have been proven to be efficient. By adjusting prices to reflect the current demand and supply situation in the market, dynamic pricing can help to prevent overstocking, improve capacity utilisation, and make markets more efficient.

But the systems that enhance market efficiency can also be used for anti-competitive results. If several participants use AI models trained on similar data, with identical goals, or models that come from the same idea of costs, then the algorithms can come up with very similar approaches to pricing. In particular, reinforcement learning models could learn that it is more valuable to emulate another competitor's price rise rather than to compete. The phenomenon is often referred to as autonomous tacit collusion – that is, when supercompetitive pricing

results without any direct communication or coordination between human actors.

Algorithmic collusion is especially problematic in markets that are concentrated because they lack competition and are marked by high informational asymmetries. The risk of convergent pricing behaviour rises when there are only a few dominant companies that share the same pricing approaches, or use similar algorithmic architectures, or have access to the same pricing inputs via third-party technologies. The Competition Commission of India (CCI), in India, has recognised these concerns in recent market studies on the e-commerce and digital platform markets. Yet, there are a few specific cases against algorithmic collusion.

This is important because it is crucial to differentiate algorithmic collusion from other coordinated behaviour. In certain instances, firms might even purposefully program AI systems to track and react to rivals' pricing patterns, thus emulating classic price-following practices. In other scenarios, coordination can be achieved organically, as the result of the independent functioning of self-learning algorithms, without the active participation of the firms involved. The two scenarios have raised many issues of competition law, but also differ in their questions of corporate governance, accountability and liability.

3. COMPETITION LAW FRAMEWORK UNDER THE COMPETITION ACT, 2002

3.1 Section 3: Anti-Competitive Agreements

The agreements prohibited by the Competition Act, 2002, are those that create or are likely to create an appreciable adverse effect on competition within India (Section 3). The word 'agreement' is used in its broadest sense, and encompasses arrangements, understandings or

actions in concert, whether or not they are intended to be legally binding or written. Importantly, Section 2(b) broadens the definition of agreement to any arrangement or understanding that can be;implied from the conduct of the parties.

Breadth is important regarding algorithmic collusion. From the judicial precedents and decisions of the Competition Commission of India, it is not always required to prove the existence of an agreement by direct evidence of communication among the parties. Parallel behaviour, market structure and other economic factors where independent behaviour is not deemed possible can be sufficient grounds to imply coordinated behaviour. This strategy is in line with the theories of tacit collusion in economics and might be especially relevant in situations where the competitors are not directly coordinating but rather through algorithms.

Section 3(3) establishes a presumption in favour of the existence of an appreciable adverse effect by certain types of horizontal agreements, including price-fixing, market allocation, bid-rigging and output restrictions. When such behaviour has been to a certain extent established, it is the responsibility of the parties involved to prove pro-competitive exceptions. This evidential framework is especially important for algorithmic collusion cases due to the lack of strict evidence of “subjective intent.” Behaviour by the same company to provide different prices for the same product in concentrated markets, plus the ability to coordinate through algorithms, could be enough to prove anti-competitive.

The primary responsibility lies with the enterprise in terms of the Competition Act. In addition, Section 48 also provides for the imposition of penalties on the officers of the company, including a director, for the contravention where they are deemed to have been responsible for or to have participated in the contravention. If enterprise liability is found, that means that directors may have secondary personal liability if they are involved in, or knew about, the wrongful act, or if they failed to prevent the wrongful act.

3.2 Abuse of Dominance under Section 4

Section 4 of the Competition Act, 2002, bans the abuse of the dominant position by enterprises in the market. AI technologies are especially pertinent in this context as they can bolster the capacity of dominant firms to engage in exclusionary/exploitative behaviours. However, AI-powered pricing systems could also allow the dominant companies to engage in predatory pricing practices by pricing their goods and services below cost to drive out competitors, which could be offset by other sources of revenue or investor financing. Likewise, algorithmic profiling can help to support discriminatory pricing, which can lead to differential pricing of consumers in an adverse manner and impact competition and consumer welfare.

AI systems could also create obstacles and create a disadvantage for smaller competitors. Algorithms may be employed to boost the ranking of their own products or services in the search results, recommendation engines, or marketplace listings for dominant digital platforms. Similarly, AI-powered loyalty programs and customised rewards can help prevent consumers from migrating to other platforms. This kind of behaviour can be considered to be exclusionary practices under Sections 4(2)(c) and 4(2)(d) of the Competition Act. Contrary to Section 3, proof of an agreement and of coordination with other market participants is not required for liability under Section 4. A single dominant enterprise can be held responsible for abusive behaviour. The difference is important to corporate boards. Directors may also be questioned on their governance responsibilities and liability if they approve, supervise or do not properly supervise AI systems that lead to abusive practices, even in cases in which they did not collude in such practices. The governance decisions of corporate boards will remain a key issue for digital markets in the future, with competition law enforcement being a battleground.

3.3 The Role of the Competition Commission of India

The Competition Commission of India (CCI) is the main competition law enforcement agency in India. The Commission has a wide range of investigative and adjudicative powers, including the power to request presentation of information, the power to investigate, the power to make market studies and the power to impose penalties. Penalties can be up to ten per cent of an enterprise's average turnover from the previous three financial years.

The CCI has taken a more proactive stance and has adopted several regulatory approaches, inspired by developments in other jurisdictions like the European Union, the United Kingdom, and the United States, in recent years. With the rise of AI-driven business models, the Commission is poised to be at the forefront of developing the legal framework for algorithmic businesses and corporate responsibility.

4. ATTRIBUTION OF ARTIFICIAL INTELLIGENCE CONDUCT

The attribution of legal responsibility for algorithmic behaviour is one of the key problems in AI-related competition lawsuits. Under the definition of an ‘enterprise', given under the Competition Act, 2002, all companies, firms, associations of persons and other economic actors involved in commercial activity fall under its scope. The Act, however, does not consider AI systems to be legal entities. An algorithm does not have a legal personality and cannot be held legally responsible. This means that any actions that stem from an AI system are considered to be the actions of the enterprise that developed, deployed, or uses the system.

This is not just a legal construct, but a reflection of the economic fact of how AI systems function in business organisations. The choice of AI systems, their goals, the data they are trained on, and the conditions in which they operate—all these are key decisions that are made by enterprises. The AI system is thus always under the control and benefit of the enterprise. Like the previous position of the Competition Commission of India (CCI), the emphasis of the Competition Commission of India is on market outcomes and their competitive impact, and not on the exact way in which coordination is achieved. For algorithmic conduct, the Commission does not have to prove that a specific anti-competitive result was an intentional objective of a human actor. Rather, it is enough to show that the result has been generated by the use of AI systems in the enterprise.

Additionally, the very nature of competition law deals with the reduction in the weight or relevance of subjective intent. Criminal law, which has been more focused on men’s rea or wrongful intent, is not so much the focus of competition law as its actual; or ;likely effects on market competition. The Competition Act, for its part, adopts concepts likeappreciable adverse effect on competition; and ;likely effect which focus on the objective market harm, and not the subjective mentality of the parties in question. An enterprise is therefore, in a normal circumstance, not able to escape from liability by claiming it didn't mean to create an anti-competitive effect. The question arises whether the enterprise's actions in designing, implementing, and controlling its AI systems foresaw conditions that would lead to anti- competitive conduct?

The same applies to the use of circumstantial evidence. The Competition Commission has always been able to infer anti-competitive agreements or coordinated behaviour from patterns of parallel conduct, market structure, economic incentives and other indirect evidence. The same type of inferential methods can be used in the realm of AI-driven markets. Coordinated conduct could be deduced even if there is no direct interaction between rivals, if multiple firms use AI models from the same provider, if they share a similar set of training data, if they have the same reward function, and if they then experience a prolonged period of supercompetitive pricing or other coordinated market outcomes.

It is therefore in line with both the nature of the law under the Competition Act and with the economic conditions of algorithmic decision-making, for the conduct of AI to be attributed to enterprises. It is important to identify such attribution to prevent gaps in accountability arising from technological innovation. If there were no principles of attribution, enterprises may be able to easily avoid liability by asserting that anti-competitive conduct was created independently by algorithms, without any human decision-making. This would be contrary to the spirit of the competition law and would further dilute the supervisory role of regulators in increasingly automated markets.

In that light, Indian competition law will likely view the use of AI systems as a means by which enterprises conduct themselves, not as an entity unto itself. The liability for algorithmic behaviour will, therefore, remain with the enterprises using systems and, if the statutory conditions are met, with the persons responsible for their operation and/or supervision.

5. CORPORATE GOVERNANCE AND DIRECTOR RESPONSIBILITIES UNDER THE COMPANIES ACT, 2013

5.1 Statutory Duties under Section 166

Section 166 of the Companies Act, 2013 provides a wide range of duties to directors. Directors must act in accordance with the articles of association, exercise their powers in good faith and have regard to the interests of the company as a whole and to the objects of the company. Directors are expected to also take into account the interests of employees, shareholders, the community and the environment when doing their duties. Moreover, they must be able to make independent decisions, act reasonably, skilfully and diligently and avoid conflicts of interest.

These responsibilities are not just 'good to do' principles, but enforceable requirements. Violations of the obligations imposed by Section 166 could result in civil liability for the directors and, in some cases, criminal liability if the violation is committed in connection with fraud or other acts of wilful misconduct or in violation of the statute. These duties are shared, but also held by each director of the board.

As an application of Artificial Intelligence (AI), deploying AI systems is a governance decision in the end, and is in the domain of the board. Accountability cannot be shifted to other people (such as a management committee) just because they have made the decision together or because the management has delegated technical functions. If a board allows an AI system to be implemented without performing proper due diligence on the risks and implications of the system, it could be questioned whether the directors have met their

statutory duties under Section 166.

5.2 Duty of Care and Informed Decision-Making

The duty of care is the duty of the directors to take care, skill and diligence as that of a person exercising the same functions. The standard is a combination of objective and subjective components. Each director is required to have the same level of care as a reasonably diligent person. But for directors who have specialized knowledge or qualifications in a particular field, they may be held to a higher standard than other directors because of their knowledge and experience in that area.

This is especially important when it comes to AI governance. For instance, if a director has background or experience in technology, data science, digital markets, and/or cybersecurity, the or she may be expected to exercise a higher level of care when reviewing an AI related

decision. Such directors might have to find and evaluate risks that a general director may not be able to see.

There's no need for every director to be an expert on machine learning or computer science. However, it must be a chain of collective knowledge on the part of the board, either as a whole or through external advisors, regarding the nature, purpose, limitations and legal

aspects of the company's AI systems. Thus, any decision to use AI must be made on reliable data and a comprehensive risk analysis of operational, regulatory and legal risks should be carried out.

Especially, in the field of competition law, it is crucial. Directors should make sure that any AI system used for pricing, market analysis or strategic decision making has been properly analysed, both legally and technically, before approval. The duty of care could be breached if the expert advice is not requested or if the possibility of foreseeable competition law issues is not investigated. In several jurisdictions, courts and regulators have increasingly spoken of the need for directors to make an adequate inquiry before approving the use of potentially high-risk technologies despite their technological complexity.

5.3 The duty of oversight and internal controls.

While the word 'duty' of oversight" is not explicitly present in Indian company law, it could be said to be a duty under the statutory provisions on internal financial controls provided in Section 166 of the Companies Act, 2013 and the respective provisions in the Listing Obligations and Disclosure Requirements (LODR) Regulations issued by the Securities and Exchange Board of India (SEBI). Together, these provisions mandate that boards have and uphold effective systems of control, monitoring and compliance across the company.

The role of oversight includes not only the decision to roll out AI systems, but also the assurance that they are doing so correctly. Boards need to ensure that they have the right governance processes in place to ensure that AI systems are performing as they should, and that they are making a positive difference. These can range from periodic reporting systems, auditing processes, escalation to abnormal results, to systems that enable meaningful human involvement and intervention when needed.

If the management alone is left to assume responsibility for the AI systems and there is no board monitoring, then directors could be questioned about their good governance. If an inadequately supervised AI system is then used to inflict anti-competitive effects, regulatory violations or other serious legal problems, the lack of adequate oversight mechanisms may come into play when determining director liability.

There are parallels with environmental, health and safety governance that can be helpful. In time, Boards have understood that they are responsible for overseeing environmental and workplace risks as part of their governance responsibilities. Likewise, legal and regulatory considerations of AI should be considered on the board's agenda. By creating a robust AI governance framework, regularly reviewing AI compliance, implementing risk assessment procedures, and reporting regularly to the board, a company can not only mitigate risk but also make it harder for anyone to claim liability against the board for actions taken with the use of AI systems.

6. BOARD LIABILITY FOR ANTI-COMPETITIVE ARTIFICIAL INTELLIGENCE

6.1 Direct Liability for Knowing Approval

The most important legal risk relates to the situation where, facing an awareness that an AI system has the potential to enable anti competitive activity, the directors knowingly approve the use of the system. A director may be found liable if they approve the use of the algorithm, if it is designed, selected or configured so that it poses a foreseeable risk of unlawful coordination, market manipulation or abuse of market power.

Section 48(1) of the Competition Act, 2002 does provide that where a company does commit a contravention of the provisions of the Act then every person in charge of and responsible for the company's business at the time of the contravention shall be deemed guilty and be subject to proceedings and be liable to be punished in accordance with the provisions of the Act with respect to such contravention provided that such person may establish a defence in accordance with the provisions of the Act. The provision mirrors elements present in other regulatory provisions and can also apply to directors who were actively involved in decision-making or who approved the business decisions.

For such a causal connection to exist between the conduct of the director and the anti- competitive outcome, it must be sufficient to give rise to direct liability. Creating such a connection can be difficult in the context of AI systems, as it can be hard to trace the multiple technical processes between algorithmic decisions and a specific market outcome. However, if evidence shows that a director knew a significant competition law risk existed with an AI system but still approved the use of the system without proper measures put in place, then the conditions for personal liability may be met.

If the director has internal reports, legal opinions, compliance warnings, or documented concerns about the competitive implications of an AI system, this may be especially relevant in determining the director's knowledge. With the growing understanding of the algorithmic risks faced by companies, regulators may be more inclined to draw inferences of anti- competitive outcomes where directors fail to pay attention to more obvious signals.

6.2 Negligence-Based Liability

In practice, it is more likely that directors will be held liable on grounds of negligence than intentional misconduct. Negligence-based liability occurs when directors do not take reasonable steps. to identify, assess and mitigate foreseeable competition law risks of AI systems.

Activities can range from approving an AI pricing system without legal counsel on competition law compliance, to not conducting an algorithmic impact assessment, failing to put in place monitoring practices, or not taking note of warning signs that an algorithm may

have anti-competitive outcomes. Under these conditions, the basis for liability is that the directors were not as careful as a reasonable and diligent director.

While negligent liability is more likely to be proved than intentional liability, it does require that the standard of care has been breached. What is a reasonable action is situational and will develop over time. The level of directors' awareness of competition law implications of AI and the extent of regulatory guidance will inevitably increase the expectations of what is required of directors.

This means that the directors who did not know of or consider algorithmic risk several years ago may be assessed differently than directors who do not consider similar risks after regulatory guidance, industry practices or enforcement actions raise awareness of algorithmic

risks. Duty of care, thus, evolves in tandem with developments in technology and regulation.

6.3 Proportionality and Personal Liability

Indian law provides for various types of personal liability. Under the Competition Act, 2002, Directors could be liable to monetary fines in certain circumstances, under the Companies Act, 2013, Directors could be disqualified in appropriate circumstances and under civil law, Directors would be exposed to civil liability for breach of fiduciary duties on behalf of the company or its shareholders. Under the Competition Commission of India, the powers are given to impose penalties on both enterprises and persons in the event of meeting the statutory requirements. Although corporate liability is always at the heart of competition law enforcement, the Commission in other regulatory fields has shown that it will also punish individuals where there is clear evidence of personal involvement and/or culpability. More importantly, personal liability is measured on a pro rata basis. The level of exposure will vary depending on the amount of involvement of the director, their knowledge and understanding of the risks, and whether they have the power to influence the decision-making process in the company. A director who takes a proactive role in picking, approving or overseeing an AI system, and who ignores warnings about problems with competition law, will be under closer scrutiny than a non-executive director who only gets a little information about the technology. But every director continues to have all the statutory obligations under the Companies Act, 2013. The failure to take an active part in management or to rely on the management alone may not be enough to provide a defence where a director had the power and the duty to take an active role in overseeing the management. Directors need to ensure sufficient governance practices, compliance programmes and risk-management systems are established to mitigate the competition law challenges of using AI.

The increasing use of AI in commercial decision-making, therefore, requires boards to adopt a proactive approach to governance. Otherwise, both companies and individual directors could face serious regulatory, monetary and reputational penalties.

7. RISK MANAGEMENT, FORESEEABILITY, AND THIRD-PARTY AI SYSTEMS

7.1 Foreseeability as a Liability Threshold

The concept of foreseeability is significant for determining the liability of directors in cases where liability can be imposed under the concept of negligence and also when determining whether an enterprise has taken reasonable measures to avoid competition law infringements.

If it is a risk that was truly unforeseeable at the time of the decision, then it shouldn't normally create any liability for the enterprise or its directors. The threshold of foreseeability is not fixed but continually shifts as there is a greater awareness within the business community, technological advances and the publication of regulatory guidance on specific risks.

Foreseeability has come into greater play in the realm of AI and competition law. Algorithmic collusion has been a topic of debate in the literature and policy fields since at least the early 2010s. Over the last few years, several international organisations have carried out in-depth research into the interaction between AI and competition law, including the Organisation for Economic Co-operation and Development (OECD) and various other regulatory bodies. Furthermore, multiple key jurisdictions have launched guidance and enforcement measures focusing on algorithmic pricing and digital market practices. The

Competition Commission of India (CCI) has also recognised the concerns with algorithmic coordination in its market studies and observations. As a result, companies in industries like technology-driven markets, financial services, e-commerce, and online platforms might have a harder time justifying that the threat of algorithmic collusion was unforeseeable.

The concept of foreseeability is not only a legal one, but also a practical one to lay down a basis on which to build a risk management strategy. Boards should take reasonable measures to detect and limit anticipated risks before installing AI systems. These can encompass algorithmic impact testing, ongoing monitoring of the impact of algorithmic pricing

decisions, audit trails on the outputs of algorithms, and escalation protocols for detecting and reacting to unusual competitive practices. Boards whose records show active involvement in these risk-management activities will be in a far better position than those that only have management or outside vendors assurances.

7.2 Third-Party AI Systems and Vendor Accountability

In recent times, an increasing number of businesses are choosing to use AI solutions provided by third-party vendors rather than developing their own in-house systems. While this approach can offer cost and efficiency advantages, it also brings additional challenges related

to governance and compliance. If multiple competitors use the same AI service provider, it could create issues similar to those that arise when competitors share pricing or information through intermediaries. For example, if different companies that compete with each other use

the same AI system to set prices, there is a risk that this could lead to coordinated market behaviour without direct communication between them. It is important to note that using third-party AI systems does not transfer legal responsibility from the company to the vendor. Under corporate and competition law principles, a company remains accountable for the actions of the tools, agents, and technologies it uses in its operations. As a result, companies cannot avoid liability by claiming that anti-competitive effects were caused by the actions of a third-party technology provider. Therefore, when evaluating AI vendors, companies should not focus solely on commercial and technical factors; they must also consider compliance with competition law. Boards and management should ask about how the AI system generates pricing recommendations, what data the AI uses to make its decisions, whether competitors are using the same system, and what safeguards are in place to prevent the AI from leading to collusive or anti-competitive outcomes.

Warranties, representations, audit rights, and indemnities related to compliance with competition law are important risk management tools. However, these provisions alone do not eliminate liability. A company that relies solely on statements from the vendor about potential risks, without conducting its own independent assessment, may be seen as failing in its duty to govern. In the end, the board should be able to show that it made an informed and independent decision regarding the use of the AI system, rather than simply deferring that decision to the service provider. This growing dependence on external AI vendors highlights the need for strong corporate governance. Directors must take reasonable steps to ensure that proper due diligence, governance, and compliance measures are in place to address all competition law concerns related to the use of third-party AI systems.

8. REGULATORY CHALLENGES AND ENFORCEMENT GAPS

The regulation of AI via competition law is fraught with numerous structural problems in India that are yet to be recognised. The first is that India doesn't have special laws concerning AI. The Digital Personal Data Protection Act, 2023, is a law on data privacy, which does not specify the behaviour of AI. There are multiple policy papers and drafts released by the government, but there is no legislation on the books regarding AI. It leads to uncertainty for companies wanting to know their legal obligations, and for regulators wanting to enforce them.

Second, a lot of corporate boards are not technically knowledgeable enough to comprehend and run complex algorithmic systems. The average age and professional history of the boards of listed companies in India is not exactly prime fodder for deep engagement with machine

learning architectures, reinforcement learning reward functions or technical roots of algorithmic coordination. The analytical ability to detect and investigate algorithmic collusion is not a given, but something that needs to be enhanced through targeted algorithmic collusion capacity-building programs for the Commission and also for the institutional shareholders of companies. The definition of;market outcome is consistent with algorithmic coordination; is not captured by the traditional tools in competition analysis, and the tools of econometrics and computation are needed to identify when the actual market outcome is consistent with such coordination, and when it is not, because of common cost conditions or demand signals. If the Commission is to take any real action in its enforcement efforts, it must develop its technical capabilities, which it can do in part by cooperating with academic institutions and international regulators. For the fourth, there is a challenge of jurisdiction and transboundary. Some of the AI systems used by the Indian companies are developed and offered in the Indian market by American, European or foreign vendors. In principle, the question of applicability of the Indian competition laws to conduct outside of India, but which nevertheless impacts Indian markets, has been answered, and it will present significant jurisdictional and enforcement challenges for foreign vendors of AI. Addressing the regulatory aspects of globally deployed AI systems will require international cooperation, including via the ICN.

9. POLICY RECOMMENDATIONS

The analysis above indicates some specific recommendations for reforms and initiatives that would make the legal landscape surrounding AI-related competition law risks in India robust.

The Competition Commission of India (CCI) should provide clear regulations on the use of the Competition Act for AI-related market practices. The Competition Commission of India (CCI) should issue guidelines for AI-driven market practices under the Competition Act. Such guidance should explain when pricing will be deemed sufficient evidence of an agreement under Section 3, when the use of a common third-party AI tool by other competitors will trigger investigation, what will be expected of enterprise and board-level governance of AI systems that have competitive implications and what will be deemed to be safe harbours or mitigating factors in enforcement decisions. This would help companies to develop compliant governance systems, and it would communicate the Commission's enforcement priorities to the market.

Second, an explicit focus on AI risk management is warranted in corporate governance standards. The Ministry of Corporate Affairs and SEBI should recommend amendments in corporate governance codes applicable to listed companies to mandate AI governance policies, board-level accountability for AI risk management and reporting of material risks arising out of AI in annual reports. This would be similar to what is happening in the EU, the UK, and the USA, and would bring Indian corporate governance in line with global norms.

Third, there is a need for capacity building within the Competition Commission. This also involves hiring specialist personnel with data science, machine learning and digital market economics expertise, and providing access to computational tools to help analyse largevolumes of pricing data to see if there is algorithmic coordination. The Commission should also explore creating an AI deployment notification system for companies to submit self- reporting of AI deployments with potential competition law challenges, in a similar way to merger notifications, to provide an early warning system.

Fourth, in the competitive marketplace, businesses should be encouraged to adopt voluntary standards for responsible AI deployment. Based on feedback received from the Competition

Commission and the Ministry of Corporate Affairs, the industry associations could come up with sector-specific codes of conduct on algorithmic pricing, transparency and governance. Enforcement actions might consider voluntary compliance with such codes as a mitigating

action. Five, the Indian government could seek global synergies in regulating AI, through bilateral dialogue with its key trading partners and multilateral forums such as the OECD, the G20 and the International Competition Network. As AI's deployment becomes a trans-border

phenomenon, there is a need for harmonised standards and measures of enforcement across the board, and India is well placed to take a positive lead in developing these global standards.

10. CONCLUSION

This application of AI in decision-making processes of the business is one of the most complex and impactful business law and corporate governance challenges that India's competition law has faced. This is because of the rise of algorithmic collusion, attribution problems, gaps in expertise within the boardrooms and regulatory bodies and the absence of specific legislation on AI. That doesn't mean, however, that no one is held accountable. The existing framework, such as the Competition Act, 2002, the Companies Act, 2013 and the principles of corporate governance for listed companies, is sufficient to hold the corporate boards accountable for their actions that will infringe the laws. By extending the definition of “agreement” in the Competition Act, its emphasis on effect instead of intention and provisions for personal liability of directors, there is an adequate enforcement structure in place. The duties of care, diligence and good faith extend to directors to have an understanding of, control over and oversight of AI systems that present legal risks. AI can't be delegated to boards, management, and vendors. Algorithmic misconduct has legal, financial, reputational and competitive implications that are board-level problems that need board-level solutions. The legal requirement for increased regulation and algorithmic impact assessments, ongoing monitoring, legal review of AI vendors and the establishment of escalation and review processes is now becoming the norm.

In the interim, the paper has identified some of the critical gaps in the current system that must be rectified through policy. The implementation of AI and competition law is not just legalistic, but also faces the challenge of insufficient regulatory clarity, technical capacity at the Competition Commission, a lack of expertise on company boards, and cross-border complexity of AI regulation. Requires legislation, regulations to invest, and industry cooperation. Artificial Intelligence doesn't take away responsibility. Rather, the market, resource, and public welfare effects of these systems put a heavier burden on the corporate companies' boards to ensure that these systems are deployed in a legal, moral and long-term beneficial manner for the company as well as society at large. It's not only about the legal obligations, but also about seeing whether or not Indian corporate governance is ready for the challenges of the digital era.

REFERENCES

Areeda, P. & Hovenkamp, H. (2018). Antitrust Law: An Analysis of Antitrust Principles and Their Application. Wolters Kluwer.

Calvano, E., Calzolari, G., Denicolò, V. & Pastorello, S. (2020). Artificial intelligence, algorithmic pricing, and collusion. American Economic Review, 110(10), 3267-3297.

Competition Commission of India. (2020). Market Study on E Commerce in India. New Delhi: CCI.

Competition Commission of India. (2022). Market Study on Telecom Sector in India. New Delhi: CCI.

Ezrachi, A. & Stucke, M.E. (2016). Virtual Competition: The Promise and Perils of the Algorithm-Driven Economy. Harvard University Press.

Ezrachi, A. & Stucke, M.E. (2017). Artificial Intelligence and Collusion: When Computers Inhibit Competition. University of Illinois Law Review, 2017(5), 1775-1810.

Harrington, J.E. (2018). Developing competition law for collusion by autonomous artificial agents. Journal of Competition Law & Economics, 14(3), 331-363.

Iacobucci, E. (2014). Tying as Quality Control: A Legal and Economic Analysis. Journal of Law and Economics, 57(2), 383-420.

Iansiti, M. & Lakhani, K.R. (2020). Competing in the Age of AI: Strategy and Leadership When Algorithms and Networks Run the World. Harvard Business Review Press.

Ministry of Corporate Affairs. (2021). Report of the Company Law Committee. Government of India.

OECD. (2017). Algorithms and Collusion: Competition Policy in the Digital Age. Paris: OECD Publishing.

OECD. (2021). Recommendation of the Council on Artificial Intelligence. Paris: OECD.

Raghavan, V. (2019). Competition Law in India: A Practical Guide. LexisNexis.

Schrepel, T. & Buterin, V. (2020). Blockchain Code as Antitrust. George Mason Law Review, 27(4), 1019-1066.

Singh, A. (2022). Competition Law and Policy in India: Cases and Materials. Eastern Book Company.

Stucke, M.E. & Grunes, A.P. (2016). Big Data and Competition Policy. Oxford University Press.

Tribunal decisions: In Re: Franchisee Association of Hyundai v. Hyundai Motor India Ltd., CCI Case No. 36 of 2014; In Re: Cartelisation in the Industrial and Automotive Bearings Market, CCI Suo Moto Case No. 05 of 2017.

Whish, R. & Bailey, D. (2021). Competition Law (10th ed.). Oxford University press