Anthropic Public Record surveys 52,000 Americans on AI

Anthropic Public Record polled 52,000 Americans via YouGov, revealing that 71% want AI regulation while only 15% trust AI companies to guide development.

To gauge AI perception beyond its direct users, Anthropic is launching Anthropic Public Record, a series of surveys, the first wave of which polled nearly 52,000 Americans via YouGov in late 2025. In terms of hopes, curing diseases like cancer or Alzheimer's clearly tops the list, ahead of assistance for people with disabilities; expectations related to companionship or therapy bring up the rear. Regarding fears, job loss dominates everywhere (64%), followed by cognitive dependence (56%) and disinformation (52%); concrete and immediate fears largely outweigh the scenario of AI "going rogue."

Two nuances emerge. Concern diminishes with use: daily users are much less apprehensive about job loss (54% versus 70% among non-users) as well as dependence. And this dependence is primarily anticipated, with only one-fifth of those concerned feeling truly disrupted if AI were to disappear overnight. On governance, the consensus is clear and bipartisan: 71% desire government involvement, prioritizing privacy, child protection, and liability in case of harm. Crucially, only 15% of respondents say they trust AI companies to decide on the technology's development, the lowest score among all institutions tested.

The company, which intends to repeat the survey before expanding it outside the U.S., aligns these results with its Advanced AI Framework: mandatory independent safety tests, transparency requirements, and the state's power to block or recall a deployment deemed dangerous. A demand whose impact it has just felt firsthand, to its own cost, since it was precisely this lever that cut off access to Fable 5 and Mythos 5.