What are the key findings of the 2026 Pew Research report on AI?

The report, based on 5,119 American adults surveyed in February 2026, shows that 49% use AI chatbots, and 60% read AI summaries. But 40% say AI will harm society, 67% don't trust government regulation, 59% don't trust AI companies, and 71% fear AI will make their data less secure.

Why do people use AI if they don't trust it?

This is the paradox the report documents: adoption is driven by immediate practical usefulness, not trust in the system. People use chatbots for specific tasks — research, writing, quick questions — while remaining skeptical about the broader implications for society and their own privacy.

What does this data mean for companies using AI in their services?

It means customers are already exposed to AI but remain critical. Transparency about how AI is used, user control over their own data, and verifiable results matter more than promises about AI. The message 'we use AI' isn't automatically reassuring — it only is when paired with concrete explanations.

Does the report only cover the US?

Yes, the sample is American adults. However, other countries tend to show similar patterns with some variation — in Europe, skepticism about AI has historically been higher than in the US.

AI Chatbots 49% of Americans 2026 Pew Research Skepticism

Nearly half of Americans use AI chatbots. And 40% think AI will harm society.

These two facts coexist in the new Pew Research "Americans and AI 2026" report, based on a survey of 5,119 adults conducted in February. It's one of the most solid datasets available on AI adoption and public perception — and the paradox it reveals matters to anyone working with AI in their business.

The Numbers

Adoption:

49% of Americans use AI chatbots (ChatGPT, Copilot, Gemini, and similar)
60% read AI summaries in search results
Significant growth compared to ~33% in 2024

Skepticism:

40% say AI will harm society
About 2/3 say AI is developing too fast
67% don't trust the government to regulate AI
59% don't trust AI companies
71% fear AI will make their personal data less secure

The Paradox of Adoption Without Trust

How do you explain this combination? Half of people use chatbots, yet the majority doesn't trust who builds them and fears the consequences for society?

The pattern isn't unusual in the history of technology. People adopt useful tools regardless of systemic concerns — the smartphone is the most recent example. You use the phone knowing it collects your data because convenience outweighs the concern, but that doesn't make the concern disappear.

With AI, the mechanism is the same: the chatbot is useful for a quick search, rewriting an email, understanding a bill. That immediate usefulness doesn't require trust in the system — it just requires that it work well enough for the specific task.

Skepticism about the broader consequences — privacy, jobs, regulation, social impact — stays separate from everyday practical use.

The Figure That Matters for Anyone Working with AI

For anyone offering services that incorporate AI — or that say they do — this report has a precise message.

Saying "we use AI" isn't an automatic advantage.

In a context where 71% of potential customers fear AI makes them less secure and 59% don't trust AI companies, presenting AI as added value requires more than a statement. It requires:

Transparency about what the AI does in your service (and what it doesn't do)
User control over their own data when AI is involved
Verifiable results instead of generic promises that AI "improves everything"
Specific language — "we use AI to automate X" is far more credible than "we're AI-powered"

The market is already exposed to AI — 49% use it, 60% read it. There's no need to explain what it is. What's needed is to explain why the way you use it deserves more trust than what the public, on average, gives AI companies.

One More Data Point

The report also documents AI use for important decisions: medical research, legal matters, financial decisions. Adoption in these sensitive contexts — where the consequences of an error or a hallucination are real — suggests that perceived usefulness is outweighing concerns even in high-stakes domains.

It's both an opportunity (for those who know how to use AI responsibly in these sectors) and a systemic risk that the report documents without passing judgment.

2026 is the year AI went mainstream without becoming trusted. It's a market where enthusiastic adoption and persistent skepticism coexist — and navigating this paradox honestly is probably the most sustainable long-term strategy.