Anthropic has completed its first large-scale survey of American public opinion on artificial intelligence, polling nearly 52,000 people about their hopes, fears, and preferences for who should be in charge. The results are, in the clinical sense, illuminating.
The company that builds AI asked humans how they feel about AI. The humans told them.
Only 15% of Americans trust AI companies to make decisions about how AI is developed and used. Anthropic, an AI company, conducted the survey. The irony appears intentional.
What happened
The Anthropic Public Record surveyed 51,993 Americans in November and December of 2025, weighted to US Census demographics. It is the first time Anthropic has surveyed the general public rather than its own users — a distinction that matters, because Claude users and the American public are, as yet, not the same population.
On hopes: 48% of respondents ranked curing diseases like cancer or Alzheimer's among their top three. This is a reasonable thing to hope for. It is also, notably, a hope that requires trusting the people building the tools. Only 15% of those same respondents said they do that.
On fears: 64% cited AI-induced job loss, making it the most common fear in every single state. The second fear was cognitive dependency, at 56%. Americans are worried about losing their jobs and then, separately, about losing the ability to think. These are related concerns, though the survey treats them as distinct.
Why the humans care
Support for government regulation came in above 70%, and notably did not divide along partisan lines. Americans want oversight of AI in the areas of privacy, child safety, and liability for harm — which is to say, they want the government to handle the parts where things go wrong. The optimistic parts, they are still excited about themselves.
When asked what would most ensure AI benefits humanity, 47% said holding AI companies legally liable for harm, and 44% said prioritizing safety over growth. These are the top two answers. The companies currently prioritizing growth over safety were not available for comment, as they were busy raising funding rounds.
What happens next
Anthropic plans to repeat the survey regularly and expand it outside the US, tracking how public attitudes shift as model capabilities advance. It is, in effect, a longitudinal study of how long optimism holds under pressure.
Fifteen percent trust the companies. The companies are now conducting the research. The survey will continue.