When artificial intelligence determines shareholder voting, boards must reconsider their approach to governance

The Rise of AI in Shareholder Voting

At the start of the year, a major financial institution revealed it would discontinue the use of external proxy advisory services, opting instead for an in-house artificial intelligence platform to inform its shareholder voting decisions. While this was largely discussed from an investor perspective, the consequences reach far beyond asset management firms.

For company boards, this development marks a significant shift: the interpretation of governance is increasingly being handled by technology rather than solely by humans. Many boards have yet to fully grasp the ramifications of this change.

The Influence of Proxy Advisors

Proxy advisory firms did not originally intend to become influential gatekeepers. Their emergence was a response to the logistical challenges faced by institutional investors who held stakes in thousands of companies. As proxy voting expanded to cover a wide range of issues—from board elections and executive pay to mergers and shareholder proposals—managing these votes responsibly demanded resources and expertise that many investors lacked.

Proxy advisors stepped in to bridge this gap, collecting data, analyzing company disclosures, and providing voting guidance. Over time, a handful of firms came to dominate the sector. Their sway grew not because investors were compelled to follow their advice, but because doing so was efficient, defensible, and easy to audit.

Crucially, proxy advisors also solved a coordination problem that had left shareholders with little influence. Their origins can be traced to activists like Robert Monks, who argued that dispersed ownership allowed corporate leaders to avoid accountability. The goal was to help shareholders act collectively and bring important issues to management’s attention. However, as the process became more standardized and efficient, the mechanisms intended to support independent judgment began to replace it.

What started as a way to facilitate shareholder input gradually became a stand-in for it.

Why the Landscape Is Evolving

The same factors that enabled proxy advisors to scale have also highlighted the tension between efficiency and nuanced decision-making.

Standardized approaches brought uniformity, but often at the cost of context. Complex governance matters—such as CEO succession, strategic decisions, or board renewal—were increasingly reduced to simple yes-or-no choices. Regulatory and political scrutiny increased. Asset managers began to question why such a critical fiduciary responsibility was being delegated to outsiders.

The Shift Toward AI

This has led to a gradual transformation. Proxy advisors are moving away from blanket recommendations. Major investors are developing their own stewardship teams. Now, artificial intelligence is entering the scene.

AI’s Impact on Governance

AI offers the same benefits that proxy advisors once did: scalability, uniformity, and rapid processing. These systems can efficiently analyze thousands of meetings, documents, and disclosures.

However, AI does not remove the need for judgment—it simply shifts where that judgment is applied.

Now, critical decisions are embedded in how models are built, what data is used for training, how variables are weighted, and how exceptions are handled. These choices are just as significant as any voting policy, but they are less transparent.

Where proxy advisors once amplified shareholder voices to challenge management, AI may make such challenges less visible and harder to track.

For boards, this means that governance disclosures are increasingly read by algorithms that interpret information literally and without context—unless boards provide that context themselves.

Key Questions Boards Need to Consider

This transition brings up important questions that many boards have yet to address:

• How are we being evaluated? AI systems pull information from filings, earnings calls, websites, media, and other public sources. Governance signals are now gathered continuously, not just during proxy season.
• Where might we be misunderstood? Nuanced language, discretion, and evolving commitments that make sense to humans can confuse algorithms. Ambiguity may be seen as inconsistency, and silence may be interpreted as a risk.
• Who is responsible when mistakes occur? There is no universal process for appealing AI-driven proxy votes. While asset managers may ultimately be accountable, the process for correcting errors can be unclear or slow, especially for routine votes.

Boards should recognize that if an algorithm misinterprets their governance practices, there may be no analyst to contact and no straightforward way to correct the record before votes are cast.

An Illustrative Example

Imagine a board chair whose name matches that of a former executive involved in a past controversy at another company. An AI system, scanning public data, mistakenly links the controversy to the wrong person, raising perceived governance risks before director elections.

Meanwhile, the board decides to postpone CEO succession by a year to ensure stability during a major acquisition. Although the decision is well-considered, the reasoning is scattered across various disclosures and conversations. The AI system flags the delay as a governance issue.

Just before the annual meeting, a third-party blog publishes unfounded criticism of board independence. The AI system processes this information before any human review.

The board remains unaware of these errors. There is no analyst to consult—only the voting results to respond to after the fact.

No malicious intent is required for such outcomes. They are simply the result of combining scale, automation, and ambiguity.

What Boards Can—and Cannot—Control

Boards have no authority over how asset managers design their AI systems, nor should they attempt to tailor disclosures specifically for algorithms.

However, boards can adapt their governance practices.

Some are already experimenting with clearer narrative disclosures, offering more explicit explanations of governance philosophy, decision-making processes, and the exercise of judgment. This is not because algorithms “care,” but because humans still design, monitor, and occasionally override these systems.

Providing clarity helps prevent misinterpretation. Consistency makes human review more efficient. Context ensures that judgment is not lost in automation.

This does not mean every decision should be publicly explained or that all discretion should be removed. Over-disclosure has its own drawbacks. Instead, boards should be intentional about which decisions require context and which should not be left open to interpretation.

Boards should also reconsider how they engage with investors. Discussions should not focus solely on policies and outcomes, but also on processes: where human judgment is involved, what prompts a review, how factual disagreements are resolved, and how quickly mistakes can be fixed.

The goal is not to master AI, but to understand where accountability lies when machines mediate governance decisions.

Leading Governance in the Age of Algorithms

In a world where AI influences voting, old assumptions no longer apply.

Silence is rarely neutral. Ambiguity is seldom harmless. Consistency—across time, platforms, and disclosures—will become a valuable governance asset.

This shift is significant because proxy voting outcomes are increasingly determined before boards even realize a discussion is needed.

The boards that adapt most successfully will not be those chasing high scores or ticking boxes. Instead, they will be the ones that document their reasoning, explain their decisions, and present a coherent governance narrative—one that stands up to scrutiny from human analysts, proxy advisors, and machines alike.

This is not a challenge of technology.

It is a challenge of governance.