AI in Boardrooms: A Global Perspective
In 2014, a significant development occurred when a venture fund in Hong Kong appointed a machine-learning program named VITAL to its board. The managing partner declared that no investment decision would be finalized without VITAL’s approval. Fast forward ten years, and we observe Abu Dhabi’s International Holding Company integrating ‘Aiden Insight’ as a non-voting AI board observer. Indian boards are beginning to embrace this trend, utilizing AI for Environmental, Social, and Governance (ESG) analysis, risk management, and compliance monitoring.
SEBI Regulations and AI Accountability
The Securities and Exchange Board of India (SEBI) has recently stipulated that entities under its regulation bear sole responsibility for the outcomes of any AI or machine learning tools they employ, whether these are developed in-house or by third parties, as per the new Regulation 16C. While AI’s influence in Indian boardrooms is undeniable, the critical question is whether the Companies Act, 2013, could legally recognize an AI system as a director, capable of being appointed, fulfilling fiduciary duties, and being held liable for any breaches.
Legal Barriers to AI Directorship
The Companies Act, 2013 presents structural challenges to AI directorship. Section 149(1) mandates that boards consist solely of individuals, a specification not made lightly. Additionally, Section 149(3) requires at least one director to reside in India for a minimum of 182 days, a condition applicable only to natural persons. Furthermore, Section 152 necessitates obtaining a Director Identification Number through Aadhaar-linked e-KYC, a requirement unattainable by software.
Section 166 of the Act underlines duties that presuppose a conscious mind. Concepts such as ‘good faith’ and ‘independent judgment’ imply a level of deliberation unique to humans, not merely outputs dictated by training data. Moreover, AI cannot inherently avoid conflicts of interest, as outlined in Section 166(4), since any bias is embedded by its creators or operators.
Criminal Liability and AI
The issue of criminal liability adds complexity. Corporate criminal responsibility in India relies on identifying the mens rea of individuals as the company’s ‘directing mind and will.’ The Supreme Court’s decision in Iridium India Telecom Ltd v Motorola Inc (2011) exemplifies this principle. An AI, lacking intent or consciousness, cannot fulfill these roles. Section 2(60) defines ‘officer in default’ in ways that do not apply to AI, as it cannot form intent or be imprisoned.
Furthermore, the business judgment rule, as applied in Miheer H Mafatlal v Mafatlal Industries Ltd (1997), requires evaluating motives and beliefs that only a natural person can possess. An AI’s lack of beliefs renders it unsuitable for this legal framework.
Accountability Through Human Oversight
If a board is influenced by AI as DKV’s board was by VITAL, Section 2(60) extends liability to any person whose advice or instructions guide the board, not the AI itself. Thus, accountability remains with those managing the AI, not the AI.
Personhood Jurisprudence in India
Indian courts have extended juristic personality to entities beyond humans, such as deities, as seen in Shiromani Gurdwara Prabandhak Committee v Som Nath Dass (2000). However, this extension always requires human agency to exercise judgment. The concept does not support AI as autonomous directors but aligns with the shadow-director structure.
More Read
Proposed Legal Adaptations
Globally, legal practices align with this view. The UK mandates that directors must be natural persons, while the EU’s AI Act insists on human oversight of high-risk AI systems. The European Parliament’s 2017 proposal for ‘electronic personhood’ was discarded to prevent shielding human accountability.
Rather than establishing AI as directors, the Ministry of Corporate Affairs (MCA) could formalize a non-voting ‘AI board observer’ role, mirroring practices like those at IHC. This should include a designated human director responsible for the AI’s mandate. Section 166 could be interpreted to impose an algorithmic oversight duty, ensuring directors who use AI analyses understand its limitations, not just its outputs. SEBI’s Regulation 16C model could be extended to listed-company boards through Listing Obligations and Disclosure Requirements (LODR).
Ultimately, the Companies Act is structured around holding natural persons accountable for their mental state and actions. As AI increasingly influences boardrooms, the focus should remain on ensuring that those who heed AI remain accountable directors.
Jenil Jain is an Associate at Chambers of Advocate Aarti Nimbalkar, Bombay High Court.
