Quinnipiac poll: large majority of Americans want independent AI safety benchmarks even if development slows

A Quinnipiac University poll conducted Sept. 24–27, 2026 finds broad public backing for externally enforced AI safety benchmarks. Respondents consistently prioritize safety over speed and show low trust in companies policing themselves.

The polling suggests bipartisan, broad-based concern that cuts across typical political lines, partly because of geopolitical competition over AI capabilities. For policymakers, the near-unanimous view that guardrails are important reduces the political risk of pursuing regulation. The remaining practical choices are about who creates the benchmarks, who enforces them, and how strict they become.

Practical implications for companies and investors

Compliance and testing carry costs. If mandatory safety standards become law or common practice, firms will likely shift resources to safety testing, documentation, and independent verification. That could compress near-term profitability for some companies. Conversely, companies that can credibly demonstrate compliance and independent verification may gain competitive advantage in a regulated environment.

Trust is a scarce asset. With three-quarters of surveyed adults distrusting industry self-regulation, third-party certification or government oversight is the likely route for credibility. Firms that move early to adopt independent benchmarks and transparent evidence of safety will be better positioned if regulation becomes widespread.

Open questions that will shape outcomes

Who designs the benchmarks? Choices include federal agencies, independent standards bodies, or multistakeholder consortia. Each path has trade-offs in technical rigor, political influence, and international alignment.

Who enforces them? Enforcement could be judicial, administrative, or market-driven. The chosen mechanism will determine the speed and uniformity of compliance.

How strict will standards be? Stricter standards raise safety assurance but increase cost and could slow innovation unevenly across firms and countries.

The poll shows a public willing to accept slower AI progress in exchange for measurable safety standards and independent verification. That public view reduces political cover for inaction and raises the prospect that regulatory or standard-setting processes will gain traction. The transition will involve concrete choices about benchmark design, enforcement mechanisms, and the balance between safety and competitive dynamics in AI development.