(Credit: Prostock-studio on Shutterstock)

In A Nutshell

A national survey of 1,200 U.S. adults found 86% are distrustful of AI results, with 42% pointing specifically to AI answers that do not show where their information came from.

People reported distrusting unattributed AI search results more than everyday frustrations like medical bills and confusing legal fine print, and three in four said the internet now feels less human than a decade ago.

A parallel poll of 800 marketing executives found 74% now treat discoverable, clearly attributed content as a top or high priority, and 91% want their content to read with a more human tone.

Most people have learned to read an AI answer with one eyebrow raised. A new national survey now puts a number on that hesitation: more than four out of five Americans say they do not fully trust what artificial intelligence tells them, and plenty keep hunting for the original source anyway.

Polling 1,200 U.S. adults, researchers found that 86% are distrustful of AI results. For 42% of them, that distrust is tied specifically to AI-generated answers that cannot show where their information came from.

That single gripe, missing attribution, outranked some of the most dependable headaches in modern life. Respondents reported distrusting unattributed AI search results more than medical bills, confusing legal fine print, and airline fees. A machine that answers with total confidence but cannot point to a source, it turns out, rattles people more than a bill they cannot decode.

Where AI Distrust Comes From

A big piece of the wariness is about who controls the pipes. Three in four respondents (75%) said they worry that what they see online is steered by a small handful of companies. Four out of five (81%) said it matters to them that online information stays openly available, rather than locked behind a paywall or owned by a few large organizations.

Given the choice, people still reach for a human. When interacting with or asking for help on a business’s website, 75% said they find humans much more helpful, while only 15% favored AI, a wide margin in favor of a real person. And folks are fairly sure of their own radar: if they suspect the entity on the other end of a chat is not a person, 56% feel confident they can tell whether they are talking to AI or a human.

Skepticism also extends to how companies use the technology in their messaging. Asked which business uses AI best in its brand voice, 61% were not sure or could not name one, and another 16% said no business uses it well at all. Only about one in four could name any company at all.

86% of Americans don’t trust AI results, according to a new survey. (Credit: Pixels Hunter/Shutterstock) Inside the AI Distrust Survey

Behind these numbers sits a two-part study. Talker Research, on behalf of WordPress VIP, ran an online poll of 1,200 general-population American adults who have internet access, fielded between April 15 and April 20, 2026. Because WordPress VIP commissioned the work, the survey carries a built-in business interest, a point worth keeping in view when weighing the results.

Researchers did not stop at everyday users. They also surveyed 800 U.S. marketing executives and digital experience experts, specifically decision-makers and chief marketing officers who buy content management and digital experience platforms. Pairing the two groups let the study line up what regular people feel about AI against what the professionals trying to reach them are planning.

A Web That Feels Less Human

A through-line in the data is a creeping sense that the internet has lost some of its warmth. Compared with a decade ago, three in four people said the web today feels less human. They also suspect machines are doing more of the talking than anyone admits: the average respondent believes 55% of their internet interactions are with AI, and 54% of their interactions on a business’s website are too.

Bots, in particular, are everywhere people look. Nearly all of those polled (92%) said they have run into AI or bots on social media, and 63% said it happens often. Patience runs thin fast. On average, it takes just 40 minutes of scrolling before people start to feel worn out by the volume of bot-made content in front of them. For a generation that spends hours a day online, hitting a wall before lunch is a short fuse.

Put those findings side by side and a clear picture emerges. Accuracy is only half of what worries people now. Many also question whether a person was ever involved at all.

Brands Facing a Trust Gap

Marketers feel the squeeze, and they know it. Three in four executives (74%) said making their website content easy to find and clearly attributed when it surfaces through AI search and answer engines is a top or high priority. Nine in 10 (91%) said it matters that their content reads with a more human tone. Nearly four in 10 (39%) admitted they are “kept up at night” by misinformation coming out of AI-generated responses, a verbatim worry that shows how high the stakes feel inside the industry.

There is also a fear of vanishing. If their content is not openly available and structured for machines to read, 69% of these professionals believe their website will become completely invisible to AI search and answer engines. That fear sits in tension with what shoppers told the same study: the people these brands want to reach still trust a human voice over a machine, even as the brands race to make their pages legible to machines first.

Leadership at the company that paid for the study framed the problem as a single, merged challenge. “Brands cannot afford to treat visibility and trust as separate things anymore,” said Steph Yiu, CEO of WordPress VIP. “More than ever, the website is where a brand provides context and earns trust.” Her colleague put the shift in starker terms. “People used to build websites for other people,” said Brian Alvey, CTO of WordPress VIP. “Now you have to build websites for AI agents acting on behalf of those people.” His warning for any business whose pages cannot be read by AI was blunt: “You don’t exist.”

Attribution has become the price of admission for trust online. People are not rejecting AI outright, and they keep using it. What they want is for the machine to show its work, the way a careful student or a good reporter would. For the brands and search engines staking their future on automated answers, that may turn out to be the hardest request to satisfy.

Survey Notes Limitations and Caveats

This was an opinion survey, not an experiment, so it captures what people say they believe and feel rather than measured behavior, and it cannot establish cause and effect. Several of the headline figures are self-reported estimates, including the belief that 55% of internet interactions and 54% of website interactions involve AI; those are perceptions, not audited counts of actual bot activity. The poll was administered online to adults who already have internet access, which can skew a sample toward more digitally active respondents. Full demographic breakdowns, margin of error, and question wording were not detailed in the materials reviewed for this article. One internal mismatch is also worth flagging: the article text and the study’s own infographic list slightly different comparison percentages for items such as medical bills and airline fees, so those side categories should be read as approximate.

Funding and Disclosures

The research was commissioned by WordPress VIP and conducted by Talker Research. Because the sponsor sells content management and digital experience products aimed at the exact problem the survey measures, namely helping brands stay visible and trustworthy in an AI-driven web, the findings serve a clear commercial interest. The two executives quoted, Steph Yiu and Brian Alvey, are the CEO and CTO of the sponsoring company.

Methodology

Fieldwork ran online between April 15 and April 20, 2026. Samples: 1,200 general-population U.S. adults with internet access, and 800 U.S. decision-makers and chief marketing officers responsible for content management and digital experience platforms. Talker Research lists its full methodology through AAPOR’s Transparency Initiative on its Process and Methodology page.