Anthropic Public Record: First Wave Survey of 51,993 Americans on AI Hopes, Fears and Regulation

Anthropic has published the first wave of Anthropic Public Record, a recurring nationally representative survey of American public opinion on AI. Fielded between November 1 and December 11, 2025 with 51,993 respondents, the study found that Americans largely agree across partisan, geographic and educational lines on two things: they want AI's benefits realized, and they want accountability from the companies building it.

What the first Anthropic Public Record found

The survey's headline results describe a public that is hopeful about AI's medical potential, worried about near-term disruption, and skeptical of the industry's stewardship.

48% of Americans ranked curing diseases like cancer or Alzheimer's among their top three hopes for AI.
64% named AI-induced job loss as a personal concern, the most common fear in every state.
71% said the government should be involved in developing and regulating AI, with bipartisan support.
Only 15% said they trust AI companies to make decisions about how AI is developed and used.

Anthropic describes the overall pattern as broad consensus rather than deep division: "Americans are eager to realize AI's promised benefits but fear the disruption it may bring, and they want accountability from the companies building it." Where disagreement appeared, the company reports, it was mostly a matter of intensity rather than direction.

How the survey was conducted

Anthropic Public Record is an online survey of the US late-teen and adult internet population, age 16 and over, resident in the 50 states, the District of Columbia, or Puerto Rico. Fieldwork was conducted by YouGov from its online panel.

The study was designed as 52 parallel state and territory samples with a target of approximately 1,000 completes per state, and approximately 500 in Montana, North Dakota, South Dakota, Wyoming, Vermont and Puerto Rico. Quotas were set within each state on age, gender, education, and race/ethnicity. The achieved sample is 51,993 completes, with state sample sizes ranging from 232 in Alaska to 1,902 in New York.

Data are weighted to be representative of the US resident population age 16 and over on state, age, gender, education, and race/ethnicity. The national margin of sampling error is ±0.6 percentage points at the 95% confidence level for a proportion of 50%. State-level margins of error range from ±2.6 points in California, New York and Texas to ±9.1 points in Wyoming. Margins of error for subgroups are larger.

What Americans hope AI will deliver

Curing disease is the single most widely shared hope for AI among Americans. Asked to choose their top three hopes from a list of 17, 48% of respondents put curing diseases like cancer or Alzheimer's in their top three — 12 percentage points ahead of the second most commonly selected item.

The full ranking of top hopes:

Curing diseases like cancer or Alzheimer's: 48%
Helping people with disabilities: 36%
Making technological progress: 23%
Making life easier in general: 23%

The lowest-ranked hopes were those involving AI as a substitute for human contact, including therapy and reducing loneliness.

What Americans fear most

Job loss, cognitive dependency and misinformation are the three most common AI-related fears in the United States. Respondents were given a list of 20 possible harms and asked to flag each one they personally worried about, then rate their worry on a five-point scale; any response of 2 ("somewhat worried") or higher counted as worried.

Job loss: 64%
Cognitive dependency — AI integration leaving people unable to think for themselves: 56%
Misinformation: 52%

The next most common concerns were criminal use and surveillance. Anthropic notes that these fears are near-term and concrete, and that each has precedent in an earlier technology — automation causing job loss, smartphones fostering dependency, social media spreading misinformation. Americans were generally more concerned about misuse of AI than about misalignment, ranking criminal use and surveillance above fears such as AI "going rogue."

Anthropic cautions that the fear figures are not directly comparable to the hope figures, because the hope question asked respondents to rank only their top three items while the fear question asked them to rate every item on the list.

For all but three of the harms listed, a majority of respondents described themselves as "not worried." However, there was no potential harm about which fewer than one in four Americans had at least some concern.

Job loss is the top fear in every state

Concern about AI displacing jobs is unusually evenly distributed across the American public. It is the top-ranked fear among Democrats (67%) and Republicans (62%), among households with children (59%) and without (66%), and in every state surveyed, from Iowa at the high end (71%) to Mississippi at the low end (57%).

Job-loss concern rises with education and falls with AI usage

Two demographic patterns stand out in the job-loss data, and they point in opposite directions.

First, concern rises with education. Americans with postgraduate degrees are nearly 10 percentage points more worried about job loss than those with a high school education or less. Anthropic interprets this as the workers whose work overlaps most closely with what AI is being asked to do being the most worried — a pattern the company says is reflected in its economic research team's analysis of the global Anthropic Interviewer study.

Second, concern falls with hands-on experience. People who use AI at work every day are notably less worried about job loss than people who do not use AI at all: 54% versus 70%. Anthropic offers several possible explanations, including that hands-on experience may build skills and fluency that let people augment rather than automate parts of their job, and that experience may also reveal AI's limitations. The company suggests the effect is likely a combination of factors.

Cognitive dependency is mostly an anticipatory fear

Cognitive dependency was the second most common fear in the survey, but the data suggest it is largely a fear of something that has not yet happened.

To test whether people are actually dependent on AI, Anthropic asked respondents how much disruption they would feel if AI became unavailable tomorrow, then compared answers between those worried about dependency and those unconcerned by it. Of the 56% of Americans who expressed some worry about dependence, only roughly one fifth would feel significant disruption if AI became unavailable. Conversely, among the 44% who do not worry about dependency, a higher share — roughly one third — would feel significant disruption.

Educators are among the occupations most worried about dependency, second only to people working in arts and design. That finding echoes Anthropic's separate qualitative study of 81,000 Claude users, in which educators were 2.5 to 3 times more likely than average to report having witnessed cognitive atrophy firsthand, presumably in their students.

As with job loss, the dependency fear falls steadily with usage: Americans who use AI daily at work are 16 percentage points less worried about dependency (46%) than those who never do (62%).

Perceived capability versus desired involvement at work

Americans rate AI's competence highest on research and lowest on service and support, and their willingness to use AI at work tracks those ratings. Respondents were given 14 workplace tasks and asked both how well they thought an AI tool could perform each task today and how much AI involvement they would prefer in their own job.

At the high end, 75% of Americans said AI was as good or better than humans at research. At the low end, 44% said AI was as good or better at service and support.

On most tasks, a majority of Americans did not want AI involved in their jobs. Even on tasks they rated AI most capable — research and data analysis — nearly half of respondents said they wanted no AI involvement in their own work. Anthropic reports that acceptance of AI involvement in the workplace moves in lockstep with perceived capability: the more competent AI is perceived to be in a domain, the more likely people are to be willing to use it.

Who the heaviest AI users are

As of late 2025, about 6% of Americans used AI every day for both work and personal purposes. Anthropic calls this group "integrated users" and treats them as a preview of what more intensive adoption looks like.

Integrated users skew young, male, urban, employed and college-educated. Nearly two thirds describe themselves as people who experiment with new technology before most others or adopt it early once they see its potential, compared with 30% of the general public. The unweighted sample of integrated users is 2,717.

Integrated users are less worried than the general public across every harm on the list, though Anthropic notes this probably reflects differences in the outlook of early adopters rather than the effect of AI use itself.

What Americans want from AI governance

Government involvement in AI has bipartisan majority support in the United States. 71% of Americans said the government should be involved in the development and regulation of AI. That figure is 79% among Democrats, 68% among Republicans and 69% among Independents. A majority in every state and territory surveyed supports government involvement, ranging from 81% in the District of Columbia to 63% in Hawaii.

Where Americans want government to act

Asked about eight specific domains and how heavily the government should be involved in each, only two drew outright majority support for more than a minimal role: privacy (56%) and child safety (52%). Liability for harm followed at 49%. National security had the narrowest partisan gap of any domain, just three points between Democrats and Republicans.

What Americans want from the industry

When asked what should happen to ensure AI is developed in humanity's interest, Americans converged on two answers:

Hold AI companies legally liable for harm: 47% chose it among their top three
Prioritize safety over growth: 44%

Independent watchdogs with real power (29%) and slowing AI development for safety (27%) followed.

The trust deficit

Only 15% of Americans said they trust AI companies to make decisions about how AI is developed and used. That was the lowest figure for any institution tested, below the federal government (20%), state and local government (19%) and international bodies (20%), and far below independent experts (43%).

Integrated users were more trusting of every institution asked about, including AI companies, and were markedly less inclined to say AI development should be slowed or stopped. Even so, they support government involvement in AI at essentially the national rate — 74% versus 71% — and across the eight governance domains tested, their preferences are nearly indistinguishable from the public's.

Anthropic's related initiatives and next steps

Anthropic positions Public Record alongside its other efforts to gather public input on AI: Anthropic Interviewer, the tool for conducting in-depth interviews at scale that powered a global qualitative study of 81,000 Claude users, and the Anthropic Economic Index, which draws on anonymized Claude usage data. Public Record marks the first time Anthropic has surveyed the general public, allowing it to reach non-users of AI.

The company also points to two recently announced policy frameworks relevant to the findings. Its Advanced AI Framework proposes mandatory independent safety testing for frontier models, transparency requirements, and government authority to block or recall dangerous AI deployments. Its Economic Policy Framework lays out how governments can prepare for AI's economic impacts — minimizing job displacement where possible and supporting workers so that AI's benefits are broadly shared.

Anthropic says Public Record will be repeated regularly, evolving in scope as new topics become more salient, so that it can track how public attitudes change as model capabilities advance and adoption deepens. The company plans to expand the survey outside the US in the future.

Limitations

Anthropic states that Public Record captures what Americans believed about AI in late 2025, and that the company is treating the first wave as a baseline. Reporting conventions for the study: all percentages are weighted; unless noted otherwise, percentages use the full segment as the denominator, with "don't know" and skipped responses remaining in the base; "worried" on the fears battery means the top four boxes of a five-point worry scale; and "as good or better" on the capability battery means the top three boxes of a five-point performance scale.