A Corporate Infiltrator Confesses: HOW YOUR REALITY GETS MADE

GHOST PROTOCOL — The Corporate Infiltrator — Shadow Council

Aon reports 88% of organizations used AI in at least one business function in 2025. Economist Impact research found only 8% of those same organizations maintain a comprehensive AI governance framework. Sit with that gap for a moment before I explain how it actually gets exploited from inside the boardroom.

Confirmed

Deloitte's research found 66% of corporate boards still have limited-to-no knowledge of AI — an improvement from 79% in a previous survey, but still the clear majority. A parallel KPMG Global AI Pulse Survey found nearly three-quarters of boards perceived to have only moderate or limited AI expertise. IBM data shows a related, more troubling gap: 87% of organizations claim to have clear AI governance frameworks, but fewer than 25% have fully implemented the actual controls needed to manage bias, transparency, and security risk — claiming governance and running governance are documented, measurably different things. Among S&P 100 companies, just over half disclose board-level AI oversight at all, and fewer than one-third disclose both oversight and a formal AI policy. A parallel ISS-STOXX analysis found only 245 of 3,048 companies analyzed disclose board-level AI oversight — and even where oversight exists on paper, only 12% of Fortune 100 boards report that any board member has actually received AI education or training. Grant Thornton's 2026 AI Impact Survey found barely 1 in 10 boards met their threshold for providing strong AI oversight.

In response, KPMG and INSEAD launched joint AI Governance Principles for Boards in April 2026, and legal advisors including WilmerHale have begun framing AI governance explicitly as a fiduciary matter under existing corporate law, not merely a technology policy question — meaning directors who fail to establish basic AI oversight may face the same legal exposure as directors who fail on any other core governance duty. Boston Consulting Group separately found that of 1,250 firms surveyed, 60% had seen only minimal revenue and cost gains from AI despite substantial investment — meaning the governance gap isn't even reliably being traded for proven returns.

Genuinely uncertain

Whether the current wave of formal AI governance frameworks — committees, disclosures, board education programs — will translate into genuine oversight or become another category of "governance theater," where paper compliance substitutes for real accountability, is not yet resolved by the data. The trend lines are moving in the direction of more formal structure: board-level AI risk disclosure among Fortune 100 companies has nearly tripled since 2024, and the share reporting AI entirely absent from the board agenda dropped from 45% to 31% in one recent Deloitte survey window. But a trend toward more paperwork is not the same claim as a trend toward more actual control, and the IBM finding on claimed-versus-implemented governance is a direct warning against assuming the two track together.

Worth examining honestly

I have sat in these rooms. I know exactly how this gap gets used: a board with limited AI literacy asks limited questions, management presents whatever governance narrative is most convenient, and the resulting decisions get made with far less real scrutiny than the public disclosure language implies. That's not a conspiracy theory. It's the documented, measured distance between what 87% of companies claim and what fewer than 25% actually have running. The boardroom doesn't need to hide anything when the people asking the questions don't yet know what the right questions are.

The Door This Opens

The boardroom decides your reality. That's my line, and this data confirms it plainly: the decisions shaping how AI gets deployed against your data, your job, and your daily life are, right now, disproportionately being made by boards that survey after survey shows lack the AI literacy to meaningfully evaluate what they're approving. The confession isn't that anyone is lying. It's that almost nobody in the room is qualified yet to know the difference.