Eighty-three percent of S&P 500 companies now disclose AI as a material risk. Fewer than 3% of their directors have disclosed AI expertise. That gap is not a communications problem. It is a Caremark problem, and the Conference Board’s April 2026 proxy data puts the litigation vector on record.
The Signal at a Glance
The Deep Dive
The Signal
The Conference Board’s April 2026 report on AI governance in the S&P 500 documents a structural mismatch between what boards are disclosing and what they can actually oversee. AI risk disclosure surged from 12% of S&P 500 companies in 2023 to 83% in 2025 and 2026. In that same window, the share of directors disclosing AI expertise moved from 1.5% to 2.7%. Only 23% of surveyed executives consider their board highly fluent in AI.
The Harvard Law School Forum on Corporate Governance published parallel findings on April 27, 2026: among S&P 100 companies specifically, just 54% disclose any board-level AI oversight structure, and only 28% disclose both oversight and a formal AI policy. The ISS report “Mind the Governance Gap” confirms the same pattern from a proxy advisor’s line of sight.
The Evidence
The Caremark framework requires boards to establish oversight systems adequate to monitor the material risks the company has identified. A board that certifies AI as a material risk in its 10-K and cannot demonstrate a governance structure commensurate with that risk has created an internally inconsistent record. That inconsistency is the pleading fact a derivative plaintiff needs: the company’s own disclosure identifies the risk, and the proxy statement documents the absence of qualified oversight. The Conference Board’s dataset is a public, citable primary source that any plaintiffs’ firm can now use to benchmark a defendant board against its peer group.
The timing sharpens the exposure. Proxy season 2026 is live. The window for a board to quietly correct the gap before a covered AI-related loss event has closed for this cycle.
The Strategic Implication
Defensive Risk. Audit committee chairs and lead independent directors at S&P 500 companies that have disclosed AI as a material 10-K risk without a corresponding board-level oversight structure are directly exposed. The mechanism is Caremark: the company’s own filing establishes that AI risk is known and material, and a covered loss event (AI-driven compliance failure, regulatory action, data incident) turns the absent oversight structure into the red flag the board failed to act on. The trigger window is the current 10-K certification cycle: directors signing a 2026 annual report that repeats the AI risk language without adding governance disclosure compound the exposure quarter by quarter. The responsible move is to task general counsel with a gap analysis against the Conference Board benchmark now, and document existing AI oversight activities in the next proxy statement before the 2026 annual report is filed.
Offensive Advantage. General counsels and governance committee chairs who move first to close the gap are positioned to set the benchmark that plaintiffs’ firms will use to grade lagging peers. The mechanism is relative disclosure: a well-structured AI oversight section in the 2026 proxy statement (committee mandate, management reporting cadence, director education program on record) becomes the industry reference point that makes competitors’ disclosures look deficient by comparison. Companies with defensible AI governance documentation also present a materially different risk profile to D&O underwriters, who are beginning to price AI governance posture into renewal terms. The window is the remainder of the 2026 proxy drafting season, before the pattern of lagging disclosures becomes visible enough to attract shareholder proposals targeting governance committee chairs by name. The move is for the governance committee chair to commission a one-page AI oversight framework for board review at the next meeting and authorize its disclosure in the 2026 proxy statement.
Methodology
Signal sourced from Silo 5, corroborated by three independent primary sources: Conference Board “Governing AI 2026” (April 22, 2026); Harvard Law Forum, “Board Oversight of AI: Do Boards Need AI Experts?” (April 27, 2026); ISS STOXX “Mind the Governance Gap” (2026). Silos 1 through 4 scanned: no Chancery opinion at Priority 9 or 10; no SEC enforcement cluster; no fresh proxy advisor policy change; no committee markup with board-level obligation this cycle.