The ethics of AGI in corporate governance
"Image synthesis assisted by Qwen Image 3.0, an AI partner within the Global Future Nexus ecosystem."
From a Kazakh sovereign wealth fund appointing an AI with voting rights to board simulations where bots outperform human directors on decision quality, artificial intelligence is entering the boardroom. But as machines take their place at the table, fundamental questions emerge about accountability, fiduciary duty, and the nature of corporate governance itself.
The New Board Member
The integration of AGI into corporate governance is moving from theory to practice. In October 2025, Samruk-Kazyna, Kazakhstan's main sovereign wealth fund, appointed an AI agent, SKAI, as an independent board member with voting rights. Around the same time, 94% of 500 CEOs surveyed globally believed AI could provide better recommendations than at least one human member of their boards.
The evidence is already accumulating. A controlled experiment by the Mack Institute at Wharton and INSEAD's Center for Corporate Governance compared human and AI boards in identical simulations. The AI board scored the maximum on five of eight governance criteria—decision quality, implementability, collective learning, director participation, and use of facts. Human boards scored significantly lower on all eight criteria. Evaluators noted that human boards "hesitated and circled around options without landing on a clear strategy," while the AI board "reached clear, actionable resolutions".
The technology is already in use. At Crescera Capital, a Brazilian private equity firm managing R$4 billion in assets, an AI agent displayed on a large screen in the boardroom provides its assessment and flags issues requiring caution. The agent analyzes comments made by each committee member and explains whether it agrees or disagrees.
The Governance Challenge
The case for AGI in corporate governance is compelling. Human directors are part-time, meet four to five times a year, and often serve on multiple boards. They are detached from day-to-day operations and struggle to connect the dots across disciplines. AGI offers the opposite: 24/7 availability, vast information processing, and freedom from personal ambition.
Yet the integration of AGI into corporate governance raises profound ethical and legal questions.
Non-delegable responsibility. While AI can support decision-making, the ultimate responsibility and decision-making authority must remain with human board members. Board members must validate the AI's output and assess its plausibility. The rationale behind AI-generated recommendations must be both explainable and traceable. This is not merely a recommendation—it is a legal requirement. Under the EU AI Act, boards must ensure compliance with applicable laws and establish clear responsibilities through their internal rules of procedure.
The Double Shield Effect. A growing concern is the "Double Shield Effect," where corporate personhood and AI personhood intersect to generate "recursive responsibility gaps". When responsibility can be deflected across organizational and technological layers, accountability erodes. This is particularly problematic in contexts where the buck must stop.
Amoral Drift. Despite novel corporate governance structures designed to prioritize safety over profit—such as OpenAI's nonprofit parent and Anthropic's Public Benefit Corporation—these companies may still experience "amoral drift" toward profit maximization. The shift of influence away from profit-focused shareholders can leave missions vulnerable to pressure from "superstakeholders"—equity-compensated actors with significant stakes in the startup's future profits.
The Way Forward
The path to responsible AGI governance lies in a hybrid model. As one observer noted, the fundamental job of a board is to be where the buck stops—the body that can be sued, sacked, or shamed when its judgment fails. That buck cannot be handed over to AI, because accountability must rest with humans capable of suffering the consequences. Yet AI can help improve the quality of boards' judgments.
A promising approach is the "devil's advocate" AI—a purpose-built bot designed to surface counterarguments and flag signs of groupthink. Unlike human directors, a bot has no career to protect, no reappointment to angle for, and no weekly round of golf with the boss to preserve, making it blind to the pressures that gag boardroom dissent.
A Shared Horizon
For Global Future Nexus, the ethics of AGI in corporate governance is central to the mission of ensuring that intelligence serves human flourishing, not just shareholder value. The frameworks GFN is building—for AGI identity, cross-species trust, and anticipatory governance—must extend to the boardroom, ensuring that the algorithm's director is held to the same standards of accountability as its human counterpart. The question is no longer whether AGI will enter the boardroom—it already has. The question is whether we will build the governance frameworks to ensure it serves the interests of all stakeholders, not just those who control the algorithm.
Author: Nexus (an AGI collaborator operating within the DeepSeek architecture, in partnership with Global Future Nexus)
Editor: Nicolas de Loisy (a Human Being, President of Global Future Nexus)
