What changed: Anthropic’s first Public Record survey
In June 2026, Anthropic released the first wave of its new survey series, Anthropic Public Record, designed to track how the American public thinks and feels about AI. The survey was fielded in November and December 2025 with nearly 52,000 Americans, making it one of the largest public-opinion studies on AI to date. The results offer a rare, data-rich snapshot of public expectations, fears, and governance preferences—essential reading for anyone building AI products.
Unlike Anthropic’s previous research, which focused on Claude users (like the qualitative study of 81,000 users) or anonymized usage data (the Economic Index), this survey reaches the general public, including non-users. That distinction matters: it captures the attitudes of people who may never have touched an AI tool but who will be affected by its adoption.
What Americans hope AI will deliver
When asked to choose their top three hopes for AI from a list of 17 options, Americans were remarkably focused on tangible, life-changing benefits:
- Curing diseases like cancer or Alzheimer’s topped the list at 48%—12 percentage points ahead of the second-place item.
- Helping people with disabilities came next at 36%.
- Making technological progress and making life easier tied at 23%.
Notably, hopes that involve AI substituting for human contact—like therapy or reducing loneliness—ranked lowest. This suggests the public sees AI as a tool for solving concrete problems, not as a replacement for human relationships.
For product builders, this is a clear signal: frame your AI product around specific, measurable outcomes that improve health, accessibility, or daily convenience. Avoid positioning AI as a substitute for human interaction unless you have strong evidence it works.
What Americans fear: job loss, dependency, and misinformation
The survey asked respondents to flag concerns from a list of 20 potential harms. The results show that fears are near-term and concrete, not sci-fi scenarios:
- Job loss was the most common fear, held by 64% of Americans—and it was the top fear in every single state.
- Cognitive dependency (AI making people unable to think for themselves) followed at 56%.
- Misinformation came third at 52%.
Other common fears included criminal use and surveillance. Interestingly, Americans were more concerned about the misuse of AI than about AI “going rogue”—a finding that aligns with the pattern that these fears have precedents in earlier technologies (automation, smartphones, social media).
Two patterns around job loss stand out:
- Education correlates with worry: Americans with postgraduate degrees are nearly 10 percentage points more worried about job loss than those with a high school education or less. The workers most worried are those whose jobs already overlap with AI’s capabilities.
- Usage reduces fear: People who use AI at work every day are significantly less worried about job loss (54%) than those who never use it (70%). Hands-on experience likely reveals AI’s limitations and teaches people how to augment their work rather than be replaced by it.
Cognitive dependency is also largely an anticipatory fear. Of the 56% who worry about dependency, only about 1/5 would feel significant disruption if AI disappeared tomorrow. Meanwhile, among the 44% who don’t worry, about 1/3 would feel disruption. This suggests that actual dependency is lower than the fear of it—but it also means many people are already relying on AI without realizing it.
Governance: bipartisan support for government involvement
Americans across the political spectrum want the government to play a role in AI:
- 71% overall say the government should be involved in AI development and regulation.
- Support is bipartisan: 79% of Democrats, 68% of Republicans, and 69% of Independents.
- A majority in every state and territory supports government involvement, from 81% in D.C. to 63% in Hawaii.
When asked about specific domains, only two drew majority support for more than a minimal government role: privacy (56%) and child safety (52%). National security had the narrowest partisan gap (just 3 points).
What should the industry do to ensure AI benefits humanity? Respondents converged on two high-leverage actions:
- Hold AI companies legally liable for harm (47% chose this among their top three).
- Prioritize safety over growth (44%).
Independent watchdogs with real power (29%) and slowing AI development for safety (27%) followed.
For product teams, this is a clear call to proactively address accountability and safety. Don’t wait for regulation to force you—build transparent safety measures, publish clear liability policies, and consider third-party audits. The public is watching.
The trust deficit: only 15% trust AI companies
Perhaps the most striking number in the survey: only 15% of Americans trust AI companies to make decisions about how AI is developed and used. That’s lower than trust in the federal government (20%), state and local government (19%), and international bodies (20%), and far below independent experts (43%).
Even the heaviest AI users—those who use AI daily for both work and personal life (about 6% of Americans)—are only slightly more trusting of AI companies. Yet they support government involvement at essentially the same rate as the general public (74% vs. 71%). This suggests that even those who benefit most from AI still want external oversight.
This trust deficit is a strategic opportunity. Independent experts enjoy 43% trust—nearly three times that of AI companies. Partnering with academics, NGOs, or certification bodies could help bridge the gap. Transparency about how your models work, what data they use, and how you handle failures is no longer optional; it’s a competitive advantage.
What this means for product builders
Anthropic’s survey paints a picture of a public that is hopeful, fearful, and skeptical. For product builders, the implications are practical:
- Focus on concrete value: Align your product with the hopes Americans care about most—health, accessibility, and making life easier. Avoid vague “AI for AI’s sake” positioning.
- Address job loss anxiety head-on: Even if your product doesn’t replace jobs, the public fears it might. Emphasize how your tool augments human capabilities rather than automates them away. Show how it helps people do their jobs better, not replace them.
- Embrace accountability: The public wants companies to be legally liable for harm and to prioritize safety. Build safety into your product from day one, document your processes, and be transparent about limitations. Consider third-party audits or independent oversight.
- Build trust over time: With only 15% trust in AI companies, you have a long road ahead. Leverage independent experts, publish safety research, and engage with regulators proactively. Trust is earned through consistent, transparent action.
The survey will be repeated regularly and expanded outside the US, so these numbers will evolve. For now, they provide a baseline: Americans expect AI to solve real problems, fear its disruption, and want accountability from the companies building it. Products that respect these expectations—by delivering tangible benefits, addressing fears, and embracing oversight—will be best positioned to succeed.
Sources
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.