Directors Duties in the Age of Agentic Artificial Intelligence

Technical Summary: Directors' Duties in the Age of Agentic Artificial Intelligence

Problem Statement
The paper addresses a critical paradox in corporate governance: the rapid, transformative acceleration of Artificial Intelligence (AI), particularly agentic AI, contrasts sharply with the slow-moving, established nature of directors' fiduciary duties. While boards are under pressure to adopt AI for efficiency, new business models, and competitiveness, such adoption frequently necessitates workforce restructuring, reduced headcount, and altered work patterns. The core problem is the "tricky conundrum" directors face in balancing the fiduciary duty to act in the "best interests of the company" (often interpreted through a shareholder primacy lens) against the interests of employees as stakeholders. There is a significant gap in literature regarding how existing statutory formulations of the best interests duty allow directors to consider employee interests specifically within the context of AI adoption decisions. Furthermore, the paper questions whether the "best interests" framework is sufficiently robust to govern the displacement of human workers by autonomous AI agents.

Methodology
The article employs a blended methodological approach combining theoretical, doctrinal, and real-world insights.

• Doctrinal Analysis: It conducts a comparative analysis of four core statutory models of the "best interests" duty found in common law jurisdictions: Shareholder Primacy, Enlightened Shareholder Value, Stakeholder Friendly, and Stakeholder Value.
• Law in Context: The paper utilizes a "law in context" approach, arguing that the substance of directors' duties is not solely defined by enforceable case law or statutes but is significantly influenced by societal expectations, soft law, and emerging norms.
• Comparative Study: It examines legislative frameworks and corporate governance codes across the United Kingdom, Canada, New Zealand, India, Australia, Ireland, and South Africa.
• Case Examination: The analysis incorporates real-world examples, including litigation risks (e.g., "AI washing"), regulatory actions (e.g., SEC, FTC), and corporate case studies (e.g., Accenture, Commonwealth Bank of Australia) to illustrate the practical application of these duties.

Key Contributions

1. Typology of Corporate Purpose Models: The paper categorizes and analyzes four distinct statutory models of the best interests duty, mapping how each permits or restricts the consideration of employee interests during AI-driven strategic decisions.
2. The "AI as Stakeholder" Inquiry: It explores the theoretical possibility of AI agents being recognized as stakeholders. While acknowledging AI is non-sentient and lacks legal personhood, the paper argues that under Team Production Theory, AI agents could conceptually be viewed as stakeholders due to their functional role in achieving corporate objectives, though this remains speculative and distinct from human employee rights.
3. The "AI Washing" Risk: The paper identifies "AI washing" (exaggerating AI capabilities or adoption) as a specific breach of the best interests duty. It argues that such deception, driven by short-term market pressures, fails the good faith test and exposes directors to legal and reputational risks, drawing parallels to "greenwashing."
4. Enforcement Limitations: It highlights the "restricted pathways" for employees to enforce their interests. The paper demonstrates that derivative actions by employees are procedurally fraught and rarely successful, rendering the best interests duty an ineffective direct legal remedy for AI-related job displacement.
5. The "Law in Context" Argument: The central contribution is the advocacy for a "law in context" approach. The paper posits that because legal norms lag behind technological change, directors must look beyond strict statutory compliance to societal expectations, reputational risks, and employee morale to truly satisfy the "best interests of the company."

Results and Findings

• Jurisdictional Variance: While all four models allow for the consideration of employee interests, none mandate that employee interests override shareholder interests.
  • Shareholder Primacy (e.g., Australia, Ireland): Directors are not obliged to consider employees but are not prohibited from doing so.
  • Enlightened Shareholder Value (e.g., UK): Directors must "have regard to" employee interests but ultimately must promote the long-term success of the company for shareholders.
  • Stakeholder Friendly (e.g., Canada, New Zealand): Directors are permitted (but not compelled) to consider broader stakeholder interests.
  • Stakeholder Value (e.g., India): Directors are required to act in the interests of employees, shareholders, and the community, though the hierarchy of these interests remains legally ambiguous.
• Insulation of Directors: Courts are generally reluctant to second-guess commercial decisions regarding AI adoption, provided directors acted in good faith and on an informed basis. This creates a "safe harbor" for boards, even when decisions lead to workforce reduction.
• Reputational and Market Consequences: The paper finds that while legal enforcement is weak, the "court of public opinion" acts as a potent corrective. Poorly managed AI transitions (e.g., lack of transparency, "tone-deaf" communications, or "AI masking") can lead to significant reputational damage, employee attrition, and loss of goodwill, which ultimately harms the company's long-term value.
• The "Great AI Goodbye": The paper observes a trend where boards are choosing to replace human workers with AI to boost profitability. However, it argues that a strict shareholder primacy lens is shortsighted, as it ignores the long-term value of employee morale, retention, and the ethical imperative of treating workers with dignity.

Significance and Claims
The paper claims that the existing legal framework of directors' duties is flexible enough to accommodate AI adoption but requires a shift in interpretation from a rigid, compliance-based view to a contextual, ethical one.

• Modest Legal Claim: The authors do not claim that current laws mandate a specific outcome (e.g., that AI adoption must be halted or that employees must be retained). Instead, they argue that the "best interests" duty is broad enough to allow directors to prioritize employee welfare, retraining, and engagement as part of a sustainable long-term strategy.
• Normative Shift: The significance lies in urging directors to move beyond "minimum compliance." The paper asserts that in the age of agentic AI, the "best interests of the company" are inextricably linked to how the company treats its workforce. Ignoring employee interests in favor of short-term AI-driven efficiencies may technically satisfy the letter of the law but fails the "law in context" test, exposing the company to social and reputational hazards.
• Practical Guidance: The paper concludes that directors should proactively engage with employees, offer reskilling opportunities, and ensure transparency. This is not merely an ethical choice but a strategic necessity to maintain the "social license to operate" and protect the company's long-term viability in an era where AI adoption is reshaping the labor market.