Anthropic Survey: 64% of Americans Fear AI Job Loss | Public Record

Anthropic has released results from what it calls the Anthropic Public Record, a new recurring survey series designed to track how Americans think and feel about AI over time. The first wave, conducted in November and December 2025 with 51,993 adults sourced through YouGov and weighted to U.S. Census benchmarks, is one of the largest single-nation AI public opinion surveys ever fielded — roughly 10 times the sample size of a typical Pew Research Center study on the subject.

The headline finding is both unsurprising and striking in its scope: AI-induced job loss is the most common fear in every single state, held by 64% of Americans. That number held across party lines — 67% of Democrats and 62% of Republicans — and ranged from 57% in Mississippi to 71% in Iowa. No other fear came close. Cognitive dependency, the worry that relying on AI will erode people’s ability to think independently, came in second at 56%, followed by misinformation at 52%.

On the hope side, 48% of respondents ranked curing diseases like cancer or Alzheimer’s among their top three expectations for AI — 12 percentage points ahead of the second-ranked hope, helping people with disabilities, at 36%. Uses like AI-powered therapy or reducing loneliness ranked at the bottom of the list.

A Trust Collapse and a Bipartisan Call for Regulation

Perhaps the most pointed finding in the survey involves trust. Only 15% of Americans say they trust AI companies to make decisions about how AI is developed and used. That number sits in awkward tension with the fact that Anthropic — itself a frontier AI lab — commissioned and published the survey. The company is effectively self-reporting data that exposes deep public skepticism toward organizations like itself. Whether that reads as genuine transparency or strategic positioning is a fair question, and readers should weigh it accordingly.

On regulation, the results showed rare bipartisan agreement: more than 70% of respondents believe the government should play a role in regulating AI. Support was 79% among Democrats, 68% among Republicans, and 69% among Independents. Privacy protections (56%), child safety (52%) and liability for harm (49%) were the areas where Americans most wanted to see government action. When asked what would most benefit humanity, holding AI companies legally accountable for harm (47%) and prioritizing safety over growth (44%) ranked highest.

The Education Paradox That Should Concern Students

For students and recent graduates, the survey’s education-level finding is the most consequential data point. Job displacement anxiety rises with education — Americans with postgraduate degrees are nearly 10 percentage points more worried about AI displacement than those with a high school diploma or less. The workers most concerned about being replaced, in other words, are the ones whose jobs most closely overlap with what AI is being asked to do: research, writing, analysis and knowledge work broadly.

This isn’t irrational. Software developers between the ages of 22 and 25 have already seen nearly a 20% decline in employment compared to their late-2022 peak, and there is a real risk that if entry-level roles disappear, the career ladder through which workers develop into senior talent gets disrupted at its foundation.

The Usage Paradox: A Concrete Prescription

The survey does offer one genuinely actionable counterweight. Americans who use AI at work every day are 16 percentage points less worried about job loss (54%) than those who don’t use AI at work at all (70%). Anthropic suggests several possible explanations — daily users may develop fluency that lets them augment rather than replace their own work, or they may gain a clearer sense of AI’s real limitations through regular exposure. Probably both are true.

This finding cuts against a trend worth noting. Gen Z’s enthusiasm for AI has been declining even as usage remains steady, with younger people increasingly questioning whether AI tools are quietly eroding their ability to think, learn and create. The Anthropic data suggest that the answer to that concern is not avoidance — it’s deliberate, critical engagement.

There’s also a classroom implication. The survey found that educators are among the occupations most worried about cognitive dependency, second only to arts and design workers. A separate Anthropic qualitative study of 81,000 Claude users found that educators are 2.5 to 3 times more likely than average to report witnessing cognitive atrophy firsthand. Students should expect professors and institutions to take AI restrictions more seriously in coming semesters, not less — and should be prepared to articulate how they are using these tools without outsourcing their thinking.

How This Survey Fits Into the Broader Research Landscape

Pew Research Center remains the dominant independent voice in AI public opinion research, but its most recent comparable U.S. survey covered just over 5,000 adults. Pew has found that 50% of Americans say they are more concerned than excited about AI in daily life, up sharply from 37% in 2021. Anthropic’s methodology — asking about specific fears rather than a single concern-versus-excitement binary — produces higher and more granular numbers, but the two sets of results aren’t directly comparable.

No other major frontier AI lab — not OpenAI, Google DeepMind, or Meta AI — runs a comparable named, recurring public opinion series. Anthropic says the Public Record will be repeated regularly and expanded outside the U.S. in future waves.

The takeaway for students is straightforward: the fear of AI displacement is real, widespread and concentrated precisely among the credentialed workers that students are training to become. The best available evidence suggests that actively learning to use AI tools — not as a shortcut, but as a skill — is the most protective response. That means engaging with these systems critically, understanding their limits, and building the kind of fluency that daily users appear to have. The data are clear that avoidance does not make the anxiety go away.

Source: Anthropic