Artificial intelligence is rapidly becoming an active participant in proxy contests. No longer solely a tool for information-gathering, AI is now influencing how institutional as well as retail investors interpret arguments, assess credibility, and ultimately cast votes. For boards, management teams, and activists, this shift introduces a new variable into already complex, high-stakes situations. 

In this white paper, Kekst CNC’s contested situations team examines how leading large language models make voting recommendations when prompted to act as proxy advisors. Drawing on contested shareholder votes from 2023 to 2025, the analysis tests how AI evaluates competing narratives, which arguments carry the greatest weight, and which sources and materials most influence recommendations. 

The findings are clear. AI does not simply replicate the frameworks of traditional proxy advisors. It applies its own logic, exhibits distinct biases, and often arrives at different conclusions. 

AI is more likely to support activists than boards and management teams – with the potential reshaping the proxy fight playing field.

AI showed a greater propensity to support activist cases for change than historical recommendations from ISS and Glass Lewis. AI gave just 37% support for issuers, compared to greater than 50% for ISS and Glass Lewis over the same period.  

Articulating a clear strategic vision and execution plan is more important than ever: the most influential factors for AI are strategy, leadership, and operational improvements.

To influence AI, issuers and activists will face a new set of ingrained tendencies that shape analyses and outcomes – in lieu of structured institutional stewardship frameworks used by proxy advisers and other passive investors. The report demonstrates that AI elevates strategic and governance arguments over more tactical or personality-driven critiques that have long been central to proxy fight communications.  

Outdated playbooks will not maximize shareholder support in the AI era. Issuers and activists must consider communications through a GEO lens. 

The report also demonstrates that AI voting recommendations rely heavily on owned content – particularly the press release, once viewed as outdated – as core inputs. They also draw extensively from a broad and uneven digital ecosystem, where volume often outweighs perceived quality. As a result, communications strategies must evolve beyond traditional media priorities to reflect how information is surfaced, aggregated, and interpreted by AI. 

The research also underscores that AI is not monolithic. Different models produce different recommendations, apply varying analytical depth, and shift outputs over time. In this environment, there is no single playbook. Success requires a disciplined approach to message development, distribution, and consistency, paired with ongoing testing and refinement. 

For companies and activists alike, the implications are immediate. AI should be treated with the same rigor as any proxy advisor. Narratives must be constructed not only to persuade investors directly, but also to be accurately interpreted and amplified by algorithmic engines that increasingly shape investor behavior. 

This report provides a data-driven foundation for navigating that reality. It outlines how AI forms its recommendations, where it sources information, and what drives its conclusions. Drawing on extensive experience in shareholder activism defense, the study offers practical communications takeaways for those preparing for contested situations in an AI-influenced landscape. 

Data infrastructure and AI analytical frameworks for this study were developed in partnership with Kekst CNC research partner Five Blocks, a digital reputation management firm specializing in how organizations and individuals are represented across Google search, Wikipedia, and AI.