Boards Are Building AI Governance Faster Than They're Training the Workers Running It

Two large surveys published within days of each other at the end of September tell a split story about enterprise AI's second half of 2026. On the executive side, KPMG's Q3 2026 Global AI Pulse — 2,131 senior leaders across 20 countries, published September 24 — found formal governance structures spreading fast: 55% of organizations now operate what KPMG calls a formal AI "harness layer" of oversight, rising to 86% among companies that have already established measurable returns. Named accountability is following the same curve, with 53% assigning AI decisions to the C-suite, including 18% who put it directly on the CEO or executive committee. Planned investment climbed too, from an average $186 million per organization in Q1 to $210 million now.

But the same KPMG data complicates the narrative of orderly maturation: only 12% of organizations consistently measure AI's value against its cost across the whole enterprise, a figure that climbs to just 48% even among companies reporting returns. In other words, roughly half of the firms that say AI is paying off still aren't systematically checking the math. KPMG also found the long-standing regional gap in scaling maturity narrowing — the Americas (64% scaling or beyond), Asia Pacific (61%) and EMEA (56%) are now 8 points apart, down from 16 points in the first quarter — suggesting governance practices are converging globally even as rigorous ROI tracking lags behind.

The workforce isn't scaling at the same rate

PwC's 2026 Global Workforce Hopes and Fears survey, drawn from 49,364 workers across 48 countries and 29 sectors and published September 29, shows usage climbing sharply from the ground up: 64% of workers report using AI on the job in the past year, up from 54%, and daily use has grown from 14% to 22%. Access to the training needed to keep up with that usage is moving the opposite direction — only 51% of workers say they can get the learning resources they need, down from 59% a year earlier. The drop is sharpest among what PwC calls the "engine room," the 56% of the workforce that does the operational work of most companies: just two in five of those workers report adequate access to training, even as their AI usage rises alongside everyone else's.

What scale demands is not simply more AI, but the accountability and coordination to run it well. — Lisa Heneghan, Global Chief Digital Officer, KPMG

That line frames the governance side of the story well, but PwC's data suggests the same discipline hasn't reached workforce development. PwC's own AI Performance Study, cited in the same survey, found that companies combining AI adoption with actual workflow redesign see 7.2 times higher AI-driven performance than those that simply deploy tools — yet most CEOs surveyed say their companies aren't yet seeing a financial return on AI investment at all. That tracks with an earlier 2026 Publicis Sapient survey of 1,550 enterprise decision-makers, which found 73% of organizations report regular or widespread AI use but only 10% consider it core to operations, with 42% of leaders admitting their organization can't yet capture the value AI is theoretically providing.

The complicating wrinkle is retention risk concentrated exactly where companies can least afford it. PwC found that "front-runner" workers — the roughly 14% with the strongest AI skills, the people organizations most need to retain as they scale agentic deployments — are also the most likely to leave, with about 29% saying they're very or extremely likely to change employers within the year. Combined with cost-of-living pressure reported by 62% of all workers surveyed, the picture is less a straightforward story of governance catching up with deployment than one where boards are building accountability structures and budgets for a workforce that is simultaneously more AI-dependent, less trained, and more prone to walking out the door with the skills still in short supply.