Anthropic: Survey of 51,993 Respondents Reveals Fears — Job Loss Worries 64 Percent of Americans

Anthropic on June 12, 2026, published the results of the first in a series of planned annual surveys on American public attitudes toward artificial intelligence. The research was conducted by YouGov on a sample of 51,993 respondents in November and December 2025, with a margin of error of ±0.6 percentage points. The goal of the series is to systematically track how public opinion on AI evolves over time, providing a foundation for better-informed decisions by both policymakers and the industry itself.

Americans’ Hopes: Curing Disease Tops the List

When asked about their greatest hopes related to AI, Americans overwhelmingly gravitated toward humanitarian and medical scenarios. 48 percent of respondents ranked curing serious diseases — cancer, Alzheimer’s, and related conditions — among their top three positive expectations for AI. In second place was helping people with disabilities at 36 percent, while general technological progress and quality-of-life improvements were selected by 23 percent of respondents.

These results suggest that the American public primarily perceives AI as a potential medical and social tool, rather than an abstract productivity technology. That insight is potentially relevant for companies and research institutions shaping the priorities of AI development and application in the coming years.

Fear of Job Loss — the Leading Concern Across All Demographics

By far the most prevalent fear in the survey is automation pressure on the labor market: 64 percent of respondents cited job loss due to AI as one of their leading concerns. The study emphasizes that this fear is consistent across all demographic categories and all political affiliations — it is the only issue on which no statistically significant differences exist between groups. Cognitive dependence on AI tools followed at 56 percent, with misinformation at 52 percent.

Geographic analysis reveals certain variations: concern about job loss is highest in Iowa, where 71 percent of respondents express it, and lowest in Mississippi at 57 percent. The survey also records a counterintuitive finding: daily users of AI tools are somewhat less worried about job loss (54 percent) compared to those who do not use AI daily (70 percent), which may suggest that hands-on experience with the tools moderates the most intense fears.

Who Should Decide How AI Is Developed?

One of the survey’s most prominent findings is a deep crisis of trust toward companies developing AI. Only 15 percent of Americans believe AI companies should independently make key decisions about AI development. By comparison, 43 percent of respondents trust independent experts, 20 percent trust the federal government, and 19 percent trust state and local governments.

This trust deficit is reflected in perceptions of accountability: 47 percent of respondents believe AI companies should face legal liability for harms caused by their systems. At the same time, 44 percent of Americans prioritize safety over development speed — a clear signal that the public is not favorable toward a scenario in which the pace of innovation accelerates at the expense of safety measures.

Regulation: Bipartisan Support for Government Oversight

Despite deep polarization in American society on nearly every issue, AI regulation emerges as a rare area of bipartisan consensus. 71 percent of Americans support government involvement in regulating AI — a view shared by 79 percent of Democrats, 68 percent of Republicans, and 69 percent of independent voters.

The regulatory priorities respondents highlight are structured around three themes: privacy protection (56 percent), protection of children (52 percent), and establishing corporate legal liability for AI harms (49 percent). These results suggest that the public is not demanding abstract “oversight” but concrete legislative interventions on specific and clearly defined areas of risk — a potentially relevant policy signal for ongoing legislative debates on AI regulation.