Anthropic Public Record: Key findings from first US wave (Nov–Dec 2025)
Key Points
- 64% worried about AI-driven job loss
- 71% support government involvement in AI regulation
- Only 15% trust AI companies to set rules
Summary
Anthropic fielded a nationally representative online survey of 51,993 Americans (Nov–Dec 2025) to measure public hopes, fears, and governance preferences around AI. Results show broad public appetite for AI benefits alongside substantial worries about job loss, cognitive dependency, and misinformation, plus strong bipartisan support for government involvement and legal accountability for companies.
Key Points
- Sample and method: n=51,993 (YouGov, weighted to US Census); state samples n=232–1,902; margins of error ±2.6–9.1 pts.
- Top hopes: curing disease (48%), helping people with disabilities (36%), technological progress / life ease (each 23%).
- Top fears: job loss (64%), cognitive dependency (56%), misinformation (52%); fears are mostly near-term and concrete.
- Job-loss patterns: concern is widespread across parties and states; higher among more-educated respondents; lower among daily AI users (54% vs 70%).
- Perceived capabilities & workplace acceptance: 75% say AI equals/exceeds humans on research; acceptance in personal jobs is lower—even where competence is rated high, many prefer no AI involvement.
- Governance and accountability: 71% want government involvement (79% Dems, 68% Republicans); most want action on privacy (56%), child safety (52%), and liability for harm (49%).
- Public preferences for industry behavior: top-lever actions are holding companies legally liable (47%) and prioritizing safety over growth (44%).
- Trust gap: only 15% trust AI companies to decide development/use; independent experts are trusted more (43%).
- Integrated users (6% of Americans) are early adopters, less worried, more trusting, but still align with public on regulatory preferences.
Practical implications for engineers
- Prioritize safety-first designs and robust testing; document safety trade-offs and fail-safes to support regulation and liability claims.
- Build privacy-preserving defaults and child-safety controls to match top public policy concerns.
- Provide clear capability limits and transparency to reduce dependency fears and manage workplace adoption.
- Design features to augment worker skills (not fully automate) and provide upskilling resources; track differential impacts by education and role.
- Prepare for legal accountability: maintain audit logs, incident-response plans, and explainability artifacts that survive regulatory scrutiny.
- Engage external independent review and public-facing communication to help close the trust gap.
Next steps
Anthropic will repeat the Public Record regularly and expand geographically; engineers should monitor evolving public priorities and align product roadmaps with governance-ready controls.