Anthropic surveys 52,000 Americans on AI hopes and fears

Anthropic released the first wave of its Public Record, a survey series that tracks how Americans think about AI over time. The first round polled nearly 52,000 people through YouGov between November and December 2025, weighted to match the US Census.

The headline finding is that public hopes and fears land in places product teams rarely hear inside their bubble. The top hope is medical, with 48% ranking curing diseases like cancer and Alzheimer's in their top three. The top fear is economic. About 64% worry about job displacement, with the number high in every state and across both parties. Cognitive dependency on AI worries 56%, and misinformation worries 52%.

On governance, 71% want the government involved in regulating AI, including 79% of Democrats and 68% of Republicans. Only 15% trust AI companies to make development decisions on their own, far below the 43% who trust independent experts and 20% who trust the federal government. Forty-seven percent think companies should face legal liability for AI harms, and 44% would put safety ahead of growth.

One interior result is worth flagging. Daily AI users report 54% job-loss anxiety, compared with 70% for non-users. Anthropic reads that as evidence that hands-on experience reduces displacement fear, though it could also mean the anxious quietly avoid the tools.

Why it matters

If you build AI products or work on policy, this is the baseline the public is actually negotiating from: broad support for regulation, low trust in labs to self-govern, and job loss as the dominant fear ahead of misinformation or safety. Plan your messaging and your liability posture around those numbers, not the version of public opinion you hear at industry events.