In 2023, 12% of S&P 500 companies disclosed AI as a material risk in their annual reports. By the end of 2025, that figure had reached 83%. A report published Wednesday confirmed what many in the IR community had quietly suspected: the disclosure is real, but the governance behind it often is not.
That gap is becoming a problem - and for IR and communications teams navigating the 2026 reporting season, it is one worth taking seriously.
What the Numbers Actually Say
The surge in AI risk disclosure looks, on the surface, like progress. Companies are acknowledging that artificial intelligence affects their operations, their competitive position, their workforce, and their regulatory exposure. They are putting it in black and white for investors to read.
But disclosure is not governance. And the data makes that distinction uncomfortably clear.
While AI as a risk factor now features in the majority of S&P 500 filings, board-level AI expertise has barely moved - ticking up from 1.5% to 2.7% of director disclosures over the same period. Technology expertise among directors rose from 20% to 51%. Cybersecurity expertise climbed from 15% to 27%. AI expertise, the very thing companies are describing as a material business risk, barely registers in the boardroom.
Meanwhile, the SEC has made its position clear. AI is a top priority in its 2026 Examination Priorities. Its Division of Examinations is scrutinising whether companies' AI-related disclosures align with actual practices. The message from the regulator is unambiguous: disclosing a risk you cannot demonstrate you are governing is not a safe harbour — it is a liability.
The Investor Expectation Has Already Shifted
This is not a future risk. Investor expectations around AI governance are already moving.
Research published in March found that 65% of U.S. institutional investors now expect companies to disclose board oversight of AI governance and ethics, and nearly half expect that oversight to be codified in committee charters or governing documents. Among S&P 100 companies, only 54% disclose board-level AI oversight, and just 28% disclose both oversight structures and formal AI policies.
That means the majority of large-cap companies are falling short of what institutional allocators already expect to see - at the same time the SEC is heightening its scrutiny of whether disclosures match reality.
Shareholder activism is following. Nine technology-sector shareholder proposals in 2025 were directly linked to AI governance inadequacies. That number will grow.
Why This Is an IR and Communications Problem, Not Just a Legal One
It would be easy to frame this as a compliance challenge and hand it to legal counsel. That would be a mistake.
The credibility of a company's investment case increasingly rests on how coherently it can explain what it is doing with AI - not what it intends to do, but what it is actually doing, who is overseeing it, and what guardrails are in place. Investors are not asking these questions to be difficult. They are asking because AI deployment decisions are now genuinely material to revenue, cost structure, risk profile, and competitive positioning.
When the annual report discloses AI as a significant risk but the board has no director with AI expertise, the messaging contradiction is visible. Sophisticated investors will notice. And in the context of an earnings call, an analyst briefing, or an AGM, being caught without a clear and credible answer to a governance question does lasting reputational damage.
The companies that are handling this well - Lockheed Martin was cited this week as a standout - are doing something straightforward: they are disclosing oversight structures, identifying relevant director expertise, and making their AI governance approach legible to investors as part of the normal reporting rhythm. There is no mystery to it. But it requires deliberate preparation.
What IR and Communications Teams Should Be Doing Now
The 2026 reporting season is the moment to close the gap between what companies say about AI and what they can demonstrate they are actually governing.
That starts with an honest audit. Which AI systems are material to operations? Who oversees them? What policies govern their use? Is that oversight visible in board committee charters and director biographies? Can that story be told clearly in an earnings call, a one-on-one investor meeting, or an AGM response?
For many companies, the answers to those questions will reveal work to be done - not at the legal level, but at the narrative level. Getting the governance right is the board's job. Getting the governance story right is an IR and communications job. Right now, both are behind where they need to be.
The investors who are asking these questions are not looking for perfection. They are looking for seriousness. They want to see that leadership understands the stakes, that oversight is real rather than performative, and that the company has a credible plan for navigating AI's role in its business.
That is exactly the kind of story IR professionals are equipped to help tell - clearly, accurately, and in a way that builds long-term trust with the capital markets.