What Americans Really Think About AI: Fear, Hope and Zero Trust in Tech Companies
Anthropic has released the first results of its Anthropic Public Record, a nationally representative survey of nearly 52,000 Americans conducted between November and December 2025. The findings paint a vivid picture of where American public opinion on AI stands right now: hopeful, anxious and deeply skeptical of the industry building it.
The most striking number in the entire survey is not about fear or hope. It is about trust. Only 15% of Americans said they trust AI companies to make decisions about how the technology is developed and used, making them the least trusted institution of all those tested. That figure sits below the federal government (20%), state and local governments (19%) and international bodies (20%), and far behind independent experts at 43%.
On the hope side, Americans are most excited about what AI could do for human health. Nearly half, 48%, ranked curing diseases like cancer or Alzheimer’s as one of their top three hopes for AI. Helping people with disabilities came in second at 36%, while making life easier generally and driving technological progress tied at 23%. Notably, using AI as a therapy tool or to reduce loneliness ranked at the bottom of the list, suggesting Americans are not yet looking to AI to replace human connection.
When it comes to fear, job loss leads by a wide margin. Sixty four percent of Americans worry that AI will displace workers, and this concern cuts cleanly across party lines. It is the number one fear among Democrats (67%) and Republicans (62%) alike, in households with and without children, and in every single state surveyed. The second most common fear was cognitive dependency at 56%, the concern that over-reliance on AI will erode people’s ability to think independently, followed by misinformation at 52%. These findings closely mirror what Anthropic found in its qualitative study of 81,000 Claude users conducted earlier through Anthropic Interviewer.
One of the more counterintuitive findings in the survey is the relationship between AI usage and fear. Americans who use AI daily at work are notably less worried about job loss (54%) compared to those who never use it (70%). The same pattern holds for cognitive dependency. People who actually work with AI tools regularly seem to develop a more grounded view of both its capabilities and its limits. American public opinion on AI, it appears, softens considerably when people have firsthand experience.
On the question of regulation, Americans are united. Over 70% say government should play a role in developing and overseeing AI, with 79% of Democrats, 68% of Republicans, and 69% of Independents in agreement. A majority in every state and territory supports government involvement. When asked where they most want action, privacy (56%) and child safety (52%) topped the list. On what industry should do, Americans prioritized holding AI companies legally liable for harm (47%) and putting safety before growth (44%).
Anthropic says the survey will be repeated regularly and expanded beyond the US, serving as a baseline for tracking how American public opinion on AI shifts as capabilities advance and adoption deepens. The company also pointed to its recently announced policy frameworks covering mandatory safety testing, transparency requirements and economic support for displaced workers.
The data is a clear signal to the industry. The public wants the benefits of AI, but on terms that include accountability, government oversight and a lot more trust than AI companies have yet managed to earn.
