When an Algorithm Reads Your Proxy and AI Is on the Ballot

AI is entering corporate governance from all sides of the table. Investors are using it to process disclosure, customize voting policies, and prepare for engagements. Proxy advisors are moving toward AI enabled, policy driven research. Shareholder activists are using it to find targets and sharpen strategic theses. Boards are being asked to explain not only how they oversee AI risk, but how they assess its opportunities and broader strategic, capital, and talent implications. At the same time, board members are increasingly using AI to assist in board oversight, a development that raises new governance questions.

AI is reshaping what companies must do to meet investor expectations, as follows.

1. AI is becoming the first reader of company disclosure. At investors and proxy advisors, AI tools are rapidly improving and analyzing disclosure before human reviewers take a closer look.
2. AI is now a subject of that disclosure. The gap between what boards say about AI oversight and what oversight looks like in practice is coming under greater focus.
3. AI has become a hook for activists. Activists have flipped the conversation from risk to missed opportunity, and AI is beginning to change how activists themselves work.
4. AI is changing how governance teams prepare. It affects the materials companies create for investor conversations and how investors prepare when they walk into an engagement.

In this white paper, we examine these trends and what companies should be thinking about ahead of the 2027 season, as AI enabled voting moves toward scale.

1. The first read of your proxy is increasingly done by a machine

AI is moving from data aggregation to voting analysis

Investors are increasingly moving beyond technology driven data aggregation toward AI powered voting analysis and stewardship. In January, J.P. Morgan Asset Management told clients it would stop using ISS and Glass Lewis for U.S. voting and rely instead on Proxy IQ, an in house AI platform. “By harnessing advanced AI, we no longer need third party data collection or voting recommendations in the U.S.,” the firm said in a client memo.[1] Three weeks later, Wells Fargo moved its voting onto a proprietary policy engine built with Broadridge as its technology partner. Norges Bank Investment Management has expanded AI use across investment research, including deploying large language models (LLMs) to screen the more than 7,000 companies in its equity portfolio for financial and ESG risks.[2] While investors retain final decision making authority, AI is increasingly shaping which companies and issues receive closer scrutiny.

At the Council of Institutional Investors’ fall 2026 conference, a number of investors indicated they are testing AI assisted proxy research and exploring models or agents to analyze ballot items, inform voting, and back test results. Some investors, such as the Florida State Board of Administration (SBA), are building proprietary AI tools that bring together issuer assessments, voting history, engagement records, and third party research.[3] New fintech providers and platform capabilities are also emerging in this space. These examples point to a broader shift toward using technology to organize evidence and support policy application, while retaining investor responsibility for the vote decision.

SEC staff have signaled openness. The day after J.P. Morgan’s announcement, Brian Daly, Director of the SEC’s Division of Investment Management, told the New York City Bar Association that agentic AI in proxy voting “isn’t science fiction, it’s a near term reality,” provided AI agents are trained, their output is reviewed by humans, and their use reflects transparency, auditability, and consistency.[4] Political pressure on the proxy advisors has, in part, accelerated the shift to custom, machine assisted voting.

AI advocates say that the technology can speed up proxy voting analysis, expanding the window between publication of the proxy statement and the shareholder meeting for internal discussion of complex voting items and engagement with companies. Historically, many investors waited until proxy advisor reports were published, typically two weeks before the shareholder meeting, to begin evaluating specific voting matters in earnest.

Proxy advisors are becoming more customized and AI driven

The proxy advisors are rebuilding around the same technology. ISS has announced PolicySteward, designed to generate automated, data driven voting recommendations with rationale text. Client testing took place in October 2026, early adopters gain access in December, and a full launch is expected in January 2027.[5] Glass Lewis is moving from a single house view toward four research perspectives, explicitly citing both diverging investor priorities and advances in AI that enable greater customization. These new perspectives are slated to underpin its research from September 2027, with its Benchmark policy continuing through the 2027 proxy season. In September 2026, Glass Lewis and Clarity AI combined to create an integrated platform spanning corporate governance, AI analytics, engagement, and proxy voting. An April 2026 Glass Lewis paper acknowledges that AI can speed proxy research and apply voting policies consistently at scale, but argues that fiduciary accountability cannot be delegated to AI and that human expertise should be built into the process rather than used only to check outputs.[6]

Other providers, some of which avoid the “proxy advisor” label, are emerging with AI enabled platforms that help institutional investor stewardship teams apply customized voting policies and quickly generate timely, company specific research. They include Broadridge Financial Solutions, Tumelo, Pennant, Proxywise AI, and Proxify. In some cases, these services have the potential to displace incumbent proxy advisors.

What machine reading means for disclosure

A human investor can provide context that may not appear in the proxy statement, such as an explanation from last year’s engagement call, a longstanding relationship with management, or the strategic rationale behind a decision. A model is more literal. It can also be more exhaustive. It may catch a changed phrase, a peer set change, a footnote that cuts against the headline narrative, or a metric that moves without explanation.

That creates a new disclosure discipline. A key point buried in a chart may not be extracted reliably. A rationale discussed privately but absent from public materials may not survive automated review. And inconsistent terminology across the proxy, 10-K, earnings materials, press releases, and engagement decks can cause different systems to construct different versions of the company’s story. Companies should consider how AI tools interpret and summarize their disclosures, including whether they capture key rationales, omit important context, or flag inconsistencies.

2. AI oversight is becoming a disclosure topic, and substance matters

AI oversight is moving onto the board agenda

Board level AI oversight disclosure has grown sharply since 2023, from roughly 15% of S&P 500 companies in 2023 to 31% in 2026, according to ISS Corporate. Among the largest companies, a majority now name a committee or the full board as responsible (54% of the S&P 100 in 2026 proxies).[7] However, only a small minority disclose any director AI training, and directors themselves are candid about the gap. PwC’s 2026 Annual Corporate Directors Survey found that 71% of directors identify AI as the leading skill their boards need to strengthen to provide more effective oversight.[8]

That gap matters because AI is not simply another technology risk. It includes both risks and large opportunities, and affects strategy, capital allocation, operations, talent, and competitive positioning.

Investors are looking beyond boilerplate language

Investors are beginning to dig deeper into what oversight looks like in action and hold companies accountable where they see a lack of oversight. For example, CalPERS’ 2026 proxy voting guidelines state that it may withhold votes from director nominees where there is evidence of “failed and/or insufficient oversight of AI related risks.” Vanguard’s stewardship report notes that boards are overseeing both AI risks and opportunities and have moved to upskill directors on AI. ISS’s Governance QualityScore now rates board AI oversight, AI skilled directors, and AI policy disclosure. However, for now, AI is more of an engagement topic than a direct vote driver. In their 2026 proxy statements, Equifax, AIG, and Truist each attributed expanded AI disclosure to conversations with shareholders.

Investors say they want less boilerplate and more specificity. Saying the board “oversees AI” isn’t sufficient. Investors are increasingly focused on several questions.

1. Who is responsible for AI oversight, and how are risks identified, monitored, and escalated?
2. Do directors have sufficient expertise to oversee AI related risks and opportunities?
3. How is AI oversight integrated into enterprise risk management and business strategy?
4. How does the board assess returns on AI investments?

AI related shareholder proposals are expanding in scope

AI related shareholder proposals continued to grow in the S&P 500 for the 2026 proxy season, with 12 proposals filed, up from six in 2025. The issues are also becoming more specific, moving beyond general AI governance toward data privacy, misinformation, workforce impacts, environmental costs, and military or dual use applications. Several proposals focused on AI’s environmental footprint, while others sought enhanced board or committee oversight of AI risks.

Support remained limited. No AI related proposal that went to a vote received majority support, with support ranging from under 1% to roughly 26%.[9]

There is no single model for AI oversight

Companies are exploring different structures for AI oversight. These include full board oversight, technology committees, distributed committee responsibilities, dedicated AI committees, and combinations of these. In its most recent proxy, Microsoft states it maintains board level oversight of AI, with its Environmental, Social, and Public Policy Committee overseeing responsible AI. It treats data center capacity and supply chain as board level risks. Salesforce revised its committee charters so that AI oversight is distributed across its five board committees. Walmart runs AI through its Technology and eCommerce Committee and extended one director’s tenure for a second year beyond its 12 year term limit to retain her technology, AI, and cybersecurity expertise. Logitech explicitly describes two sided oversight, with the full board reviewing AI “from both an opportunity and risk perspective.” Groupon’s most recent proxy states it created a standalone AI Committee chaired by a newly appointed AI company founder.

Board expertise is becoming an investor focus

Board expertise continues to be an important investor focus area. A skills matrix that newly labels a director as “AI experienced” is less persuasive if the appointment record does not show meaningful technology or data leadership. The strongest model may combine director education, real operator level technology expertise, and access to outside specialists, with the approach clearly described in the proxy.

Much of the reported growth in AI expertise is likely relabeling, not new seats. 37% of S&P 500 companies now cite AI experience for at least one director, up from 11% in 2022.[10] Investors are looking at this and surfacing these changes with the help of AI.

Where boards are adding technology and AI expertise, they are increasingly turning to experienced operators, including senior data and technology executives. For example, J.P. Morgan’s Chief Data & Analytics Officer was appointed to the CVS Health board (effective November 2026) and CVS Health’s Chief Data & Analytics Officer joined the Choice Hotels board, alongside product leaders from large technology companies. AI expert seats are also emerging in activist adjacent board changes. Sysco’s board additions, announced in August, followed introductions facilitated by the D. E. Shaw group and came alongside a technology committee renamed the “Artificial Intelligence Transformation & Technology Committee.”

AI is beginning to enter executive compensation

AI metrics in executive compensation programs are still relatively rare. Mercer found that among 145 early S&P 500 filers in 2026, 14% use an explicit AI metric in short term incentives and just 2% in long term incentives.[11] Among those, IBM and Salesforce noted in their most recent proxy statements that they have incorporated AI metrics into their executive compensation programs.

3. AI is becoming an activist thesis and changing how activists work

AI is becoming a source of activist pressure

In the first half of 2026, AI featured in roughly one in six U.S. activist campaigns and nearly 30% of campaigns in the technology sector.[12] The recurring activist criticisms are that companies are not moving fast enough on AI or communicating their AI story well. Activists continue to push companies to develop AI strategies aimed at unlocking value for shareholders.

Starboard told Tripadvisor in February that “despite the risk of AI disintermediation increasing daily,” the company had been “massively late and incredibly slow” to launch a compelling AI enabled product. Elliott has pursued the AI theme at multiple companies in 2026. It reportedly pressed London Stock Exchange Group to explain its AI strategy and to counter concerns that large language models could bypass its data business. It took a stake in Synopsys on the view that AI is driving greater investment and engineering complexity in chip design, and gained a board seat in May. In June, Elliott placed a partner on Hewlett Packard Enterprise’s board under a 2025 cooperation agreement, arguing the company stands to gain from the buildout of AI infrastructure. Irenic Capital pushed Snap to cut costs and deploy AI to monetize its data. Spruce Point argued the market undervalued Zoom’s AI pivot, and Starboard urged Riot Platforms to convert its Texas power capacity from bitcoin mining to AI data centers.

AI can accelerate activists’ tactics

The second order effect is on the activists themselves. AI can accelerate target screening, peer analysis, document review, thesis development, and campaign material creation. Established activists are investing in internal data and AI capabilities. For example, in May 2026, Elliott hired a former Blackstone data science executive to run its global research and data science group, which includes roughly 40 research staff and three data scientists with a mandate to build out AI capabilities across its work.[13]

Campaigns that were once too costly could become viable for more funds. AI enables activists to screen more targets, run more campaigns simultaneously, and move faster from initial thesis to public pressure. A fight deck that once took weeks can now be drafted in minutes, though with serious limitations. For boards, that compresses the time available to discover and correct a vulnerability before an activist identifies it. Companies should be doing the same work proactively.

AI is becoming both a risk and an opportunity for boards

For boards, the question is no longer only, “Are we managing the risks?” but, “Can we explain and defend our AI strategy relative to peers, in terms an activist could not improve on?” The question is not simply “Did we disclose AI?” but whether an investor, activist, or voting system can look at the public record and understand how AI affects the company’s strategy, decisions, and governance. Boards should be preparing for that higher standard in 2027.

4. AI is changing the engagement on both sides of the table

Investors are arriving better prepared

AI is benefiting the investor on the other side of the engagement table. Investors can ingest more of a company’s disclosure history before a meeting and compare it with peers, voting guidelines, prior engagements, and public activist arguments. The result is a more informed engagement, with investors able to prepare more efficiently and enter discussions with a deeper understanding of the company and the issues at hand.

AI is well suited to discrete tasks, such as extracting disclosures, comparing peers, and identifying questions, but it should inform rather than replace human judgment. As AI assisted research becomes available earlier in the proxy process, investors may no longer need to wait for a proxy advisor report before engaging. This could expand both the window for shareholder dialogue and the overall volume of engagement, giving companies more opportunities to engage with investors, address concerns, and influence voting decisions before votes are cast.

Companies can use AI to see themselves as investors do

The same tools that investors use to analyze a company can let a company see itself as investors might. Companies can use AI to pressure test their own disclosures, identify inconsistencies across public materials, anticipate questions, and compare their positioning with peers and investor expectations.

5. Directors’ use of AI raises a governance question of its own

AI can improve board preparation by helping directors synthesize complex materials, identify emerging risks, and develop better questions. But public LLMs can expose confidential information and create security, privacy, privilege, and recordkeeping concerns. AI outputs may also be inaccurate or biased, making human judgment and verification essential. Its use in the boardroom therefore raises broader questions about oversight and accountability. The goal should not be to prohibit AI assistance, but to establish a secure framework for using it responsibly. One approach is for the corporate secretary to use approved tools to generate AI assisted summaries alongside the underlying board materials. Ultimately, the challenge for boards is to harness the benefits of AI while maintaining appropriate safeguards, oversight, and accountability.

What boards and companies should do before the 2027 season

1. Write for both readers. If material data appears only in charts, include it in the text as well; explain pay decisions and exceptions, and use consistent terminology across disclosures and engagement materials.

2. Disclose how AI is overseen. Identify who owns AI oversight, how responsibilities are divided, and how issues are reported, escalated, and measured.

3. Match expertise to the appointment record. Go beyond relabeled skills matrices by demonstrating genuine AI operating expertise, education, or a credible expert bench.

4. Pressure test the strategy against an activist. Identify potential gaps in speed, monetization, productivity, data, infrastructure, capital allocation, and disclosure.

5. Establish rules for the board’s own AI use. Define approved tools, prohibited information, human review, verification, and escalation.

6. Re map the vote. Identify holders using AI enabled voting systems, custom policies, or pass through voting arrangements. Historical voting patterns may be less predictive.

7. Engage shareholders before issues emerge. Proactively engage investors to understand evolving expectations and address potential vulnerabilities.

8. Start before the proxy cycle. Preparation for 2027 is already underway. The goal is not more sophisticated sounding disclosure, but more defensible governance and strategy.

---

1J.P. Morgan Asset & Wealth Management client memo, as reported in ESG Dive, “JPMorgan eschews proxy advisers for internal AI tool,” January 7, 2026.(go back)

2Norges Bank Investment Management, Responsible Investment 2025, February 26, 2026; Reuters, “Norway’s wealth fund using AI to screen for ESG risks,” February 26, 2026.(go back)

3Florida State Board of Administration, Investment Advisory Council meeting transcript, December 15, 2025 (included in the March 30, 2026 meeting materials).(go back)

4Brian Daly, SEC Division of Investment Management, “(Re)Empowering Fiduciaries in Proxy Voting,” remarks to the New York City Bar Association, January 8, 2026.(go back)

5ISS STOXX, press releases, April 29, 2026 and September 30, 2026.(go back)

6Glass Lewis, AI and the Fiduciary Test, April 2026.(go back)

7ISS Corporate, “AI Governance Appears on Corporate Radar,” 2024; “AI in Focus in 2025,” March 19, 2025; “2026 U.S. Governance Post Season Review,” September 24, 2026; Glass Lewis, “US AI Oversight Through Three Lenses,” February 26, 2026.(go back)

8PwC, 2026 Annual Corporate Directors Survey, September 2026.(go back)

9Diligent Market Intelligence, Insightia voting data for the S&P 500 as of October 2, 2026, covering the 2025 and 2026 proxy seasons (July 1 through June 30).(go back)

10EY Center for Board Matters, “Seven questions boards should ask after the 2026 proxy season,” August 10, 2026.(go back)

11Mercer, “Mercer Identifies a Baseline for AI Adoption in Executive Incentive Plans,” 2026.(go back)

12Skadden, “Activism Update. Fewer Proxy Contests, More AI Focused Themes,” The Informed Board, September 2026.(go back)

13Bloomberg, “Elliott Ramps Up Its AI Efforts With Key Hire From Blackstone,” May 20, 2026.(go back)

Legal Disclaimer:

EIN Presswire provides this news content "as is" without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, kindly contact the author above.