At most a quarter of the small US firms using generative AI have a plan governing it

Two of the figures describe the same population, and they only fit together one way. If 22 firms in every 100 have a governance plan and 84 in every 100 are already using generative AI ^[1][3], then at most 22 of those 84 users have written anything down, which is about 26 percent of them ^[11]. At least 62 firms in every 100 are running the tools with nothing behind them ^[12]. Use outnumbers governance by roughly 3.8 to 1 ^[13].

The column that carries these numbers is a course-correction argument, and its five strategies are stop waiting for the calendar, start with customers, uncover the real problem, build room for a surprise, and check your ego ^[14], and none of those is writing an AI policy. The AI figures sit in the setup, alongside the claim that CEOs who push out hard decisions invariably face the same situation the following year ^[6]. "The noise around AI is deafening, and CEOs who keep kicking that strategy down the road will regret it," the contributor wrote ^[7].

The test the column offers for a failing program applies awkwardly here. "The only question that matters is whether continuing will produce a different result," the contributor wrote of sunk cost ^[8]. That question works on a product line or an underperforming leader. It works less well on generative AI in these firms, where adoption is reported as already started ^[1]. The decision in front of most of them is what to permit.

I would want to know what the survey counted before acting on these figures. The column does not disclose Vistage's sample size, its field dates, or what the survey treated as a governance plan ^[10]. A signed acceptable-use policy and a slide saying someone will draft one both plausibly land inside 22 percent.

The 22 percent measures whether anyone in the firm has decided who decides. It does not measure whether any plan works. And 76 percent of chief executives are already users of the thing being decided ^[2][3]. The column also names two other pressures landing in the same weeks, without ranking them against the AI question: wages ticking back up, which makes retaining good people more expensive ^[4], and customers wanting clearer justification of the value they are getting, especially where prices have risen ^[5].

The piece runs under Entrepreneur's contributor terms, which state that opinions expressed are the contributor's own ^[9]. The 84, the 76 and the 22 are Vistage's ^[1][2][3].