Sixty-nine percent support belongs to 849 adults who received one version of one survey question. Verasight asked whether respondents supported legislation from "a Senator" requiring artificial-intelligence companies to transfer 50 percent of their stock to a fund owned by the American public. Twenty-one percent strongly supported it and 48 percent somewhat supported it. The result is Q16's unnamed-sponsor split, not all 1,690 respondents and not all American workers. [1]
CNBC described a majority of workers supporting an AI wealth fund and wrote that the national survey of 1,690 adults found 69 percent support. The direct questionnaire supplies the missing denominator: Verasight surveyed United States adults aged 18 or older, then randomized them between two sponsor treatments. CNBC's emphasis on layoffs may explain why the result is news; it does not turn an adult sample into a worker-only sample. [1][2]
The other 841 respondents saw otherwise identical wording that named Senator Bernie Sanders. In that Q15 split, 24 percent strongly supported and 40 percent somewhat supported the proposal, for 64 percent total support. The five-point difference is evidence that sponsor wording can matter inside this survey. It is not, without an uncertainty estimate for each split and a formal comparison, proof of the size of a durable Sanders effect. [1]
Verasight collected responses on June 18 and 19. It says the full sample was weighted to demographic, geographic, partisanship and 2024-vote benchmarks, and reports an overall margin of sampling error of plus or minus 2.8 percentage points. That overall figure should not automatically be assigned to each half sample or subgroup. Recruitment, weighting, question order and compensation also belong in any judgment about what the poll can represent. [1]
The instrument measures a reaction to a stark proposal: make AI companies transfer half their stock to a publicly owned fund. It does not ask respondents to design valuation, compensation, voting rights, covered-company thresholds, fund governance or distributions. Nor does it establish constitutional judgment, policy knowledge, intensity over time or what support would survive a legislative campaign naming particular firms and costs.
Question context matters as well. Respondents encountered the stock-transfer item after earlier questions about AI use, anxiety, regulation, safety and personal finances. The report publishes that order, allowing readers to inspect it, but this single survey cannot reveal how an isolated question or alternative wording would change the response.
Those implementation questions belong to the day's Sanders companion. This article owns measurement: population, question, randomized sponsor, weighting and response categories. The companion owns whether a bill exists, how an equity transfer would work and who would govern the resulting stake. A poll answer is not legislation, and a politician's argument for public ownership is not evidence for the poll's reported percentage.
No qualifying pre-cutoff X status survived the recorded poll and CNBC searches. A July 13 status accurately summarized the article but fell after this edition's cutoff, while the verified Sanders status is allocated only to the ownership-proposal story and does not prove the survey result. The empty stack therefore marks a bounded search outcome, not public silence.
The 69 percent finding is substantial enough without inflating it. It says that 69 percent of one randomized 849-adult split supported one unnamed-senator stock-transfer proposal on two June dates. Calling that result "all workers" removes the very features that let readers evaluate it. The questionnaire is not an inconvenient footnote to the headline; it is the finding.
-- THEO KAPLAN, San Francisco