Last updated October 2026. Figures throughout reflect the Fractl and Search Engine Land Q2 2026 consumer survey and Pew Research Center’s June 2026 “Americans and AI” report.
Every AI adoption chart tells the same story: more people, more often, asking ChatGPT, Perplexity, and Google AI Overviews instead of typing a search query. That chart is real, and most brand visibility teams have built their whole reporting stack around it.
A second chart, published by the same researchers, tells a harder story. The people using AI search more also trust it less than they did a year ago. Those two lines used to move together: more use meant more trust, and more trust meant more use. In 2026, they split.
Call it the AI search trust recession: a widening gap between how much people use AI search and how much they trust what it tells them. It is the single most important context a brand visibility program can have right now, because a rising mention count means something different in a year when the reader on the other end trusts the channel 28 points less than they did twelve months ago.
The ‘more helpful than Google’ number fell from 82% to 54%
In Q2 2025, Fractl and Search Engine Land asked 1,008 U.S. consumers a simple question: is AI search more helpful than a traditional search engine like Google? Eighty-two percent said yes.
One year later, the same question and the same survey design produced a different answer. Only 54% of consumers now call AI search more helpful than Google. That is a 28-point drop in twelve months, one of the sharpest reversals in recent consumer sentiment research.
The drop did not happen because people stopped using AI search. It happened because they started distrusting what it gave them.
AI skeptics grew 6x, from 3% to 17%
The flip side of a falling helpfulness score is a rising skeptic count. In 2025, only 3% of consumers said AI search was less helpful than Google. By Q2 2026, that share had grown to 17%, a sixfold increase in one year.
A skeptic is not just someone with a lukewarm opinion. A skeptic actively rates AI search as worse than the tool it is supposed to be replacing. Six times as many consumers now hold that view than did twelve months earlier.
For your brand, this matters because a skeptic does not just distrust the engine. They carry that distrust into how they read whatever the engine tells them about you.
The year-over-year trust scoreboard
Put the full Fractl and Search Engine Land data set side by side, and the trust recession stops looking like one bad data point. It is a pattern across four separate questions, all measured the same way, one year apart.
| Metric | 2025 | 2026 | Source |
|---|---|---|---|
| Say AI search is more helpful than Google | 82% | 54% | Fractl and Search Engine Land, Q2 2026 |
| Say AI search is less helpful than Google | 3% | 17% | Fractl and Search Engine Land, Q2 2026 |
| Say heavy AI use in brand marketing lowers trust | 20% | 39% | Fractl and Search Engine Land, Q2 2026 |
| Marketers who monitor their brand’s presence in LLM answers | 22% | 49% | Fractl and Search Engine Land, Q2 2026 |
Three of those four lines move against you. Only the fourth, marketer monitoring, moves in a direction that helps. More marketing teams are watching what AI engines say about their brand. That is a start, not a fix.
Adoption and trust just crossed
Here is the part that makes this a genuine paradox and not just a bad year for AI search. Usage did not fall alongside trust. It kept climbing.
The same Q2 2026 survey found 70% of consumers say they use AI search more than they did a year before. Only 3% pulled back. Pew Research Center’s June 2026 “Americans and AI” report puts a second number on the same trend: 49% of U.S. adults now use AI chatbots, and 60% say they read AI-generated summaries inside their regular search results.
THE DIVERGENCE, 2025 vs. 2026 (Fractl and Search Engine Land, Q2 2026)
"AI search is more helpful than Google"
2025 ================================= 82%
2026 ====================== 54%
"AI search is less helpful than Google" (skeptics)
2025 = 3%
2026 ======= 17%
"I use AI search more than I did a year ago"
2026 ============================ 70%Two years ago, helpfulness and adoption moved together. People used AI search more because they trusted it more. In 2026, that link broke. People are using AI search more even though they trust it less than before.
That gap is the trust recession. It does not show up in a raw usage chart, a session count, or a share of voice number. It shows up only when you ask people what they think of what they are using.
Google still wins 5 of 6 search categories
The 54% headline number hides a more specific pattern. When Fractl and Search Engine Land broke helpfulness down by category, Google kept its lead in nearly every one.
| Category | Consumers who prefer Google over AI search |
|---|---|
| Local businesses | 74% |
| Product research | 58% |
| Travel planning | 57% |
| Health | 55% |
AI wins exactly one category outright, how-to guides, and even there it shares the win with YouTube, which takes the top spot at 50%.
Health is the closest race. ChatGPT ranks as the number two destination for health questions at 26%, ahead of WebMD-class sites, even though Google still wins the category at 55%. That is the trust recession in miniature. AI has become a credible second option almost everywhere and the outright leader almost nowhere.
That per-category split matters for how you read your own AI visibility numbers. A brand can look strong in an aggregate share of voice score while sitting inside a category where the reader still defaults to trusting Google first and treats the AI answer as a second opinion to verify.
Heavy AI use in your marketing now costs you trust
The trust recession is not only about the AI engines themselves. It is spilling into how consumers judge the brands that lean on AI to reach them.
In the same Q2 2026 survey, 39% of consumers said heavy AI use in a brand’s marketing lowers their trust in that brand, nearly double the 20% who said the same in 2025.
Gen Z consumers apply the harshest standard. More than half, 54%, said heavy AI use in marketing would cost a brand their trust.
| Generation | Say heavy AI use in marketing lowers brand trust |
|---|---|
| Gen Z | 54% |
| Gen X | 33% |
| Baby boomers | 32% |
Source: Fractl and Search Engine Land, Q2 2026.
That is not a reason to hide your use of AI. It is a reason to watch how visible and how heavy-handed it looks to the audience judging you.
Share of voice alone will not catch a trust recession
Most AI visibility programs are built around one number: share of voice, the percentage of relevant prompts in which your brand gets mentioned versus your competitors (see the full breakdown in the AI visibility glossary). It is a real metric, and on its own it is the wrong one to watch in a year like this.
Share of voice can rise while trust falls. Your brand can appear in more answers, more often, while the consumer reading those answers grows more skeptical of the platform delivering them and of how it frames you. A high mention count next to a skeptical reader is not a win. It is exposure to a reader primed to doubt what they read.
The number of marketers actively monitoring their brand’s LLM presence more than doubled in a year, from 22% to 49%. Even the people paid to watch this closely are realizing that presence alone does not tell the full story. What they are adding is sentiment: whether the AI engine describes your brand positively, neutrally, or negatively, and whether that framing is shifting as the platform’s own credibility shifts under it.
What to track now: perception, not just presence
A share of voice dashboard tells you whether you showed up. A sentiment and perception layer tells you what happened once you did. In a year when trust in the underlying channel just dropped 28 points, that second layer is the one that protects you.
Evertune and Scrunch AI both build for that depth. Evertune runs a dual-layer read, foundational model knowledge alongside real consumer app responses, so you can see whether the framing your brand gets in a lab test matches what a live user actually sees. Scrunch pairs multi-engine sentiment tracking with SOC 2 Type II compliance and an Agent Experience Platform that serves AI-optimized content directly to LLM crawlers, useful for teams trying to correct a negative frame at the source rather than just watching it happen.
For teams that want share of voice, sentiment, and a fix queue in one flat-rate subscription, Temso tracks brand mentions and sentiment across ChatGPT, Perplexity, Google AI Overviews, Gemini, and Microsoft Copilot from $89/mo, and its built-in AI SEO agent can execute the fixes it surfaces. Peec AI is a credible option too, particularly for agencies that need unlimited seats across client accounts without per-user fees.
Whichever platform you choose, the questions stay the same. Is sentiment moving, not just mention count? Is your framing holding steady as the engine’s own credibility with consumers erodes? And are you checking this weekly, not quarterly, in a year when the baseline is shifting by double-digit points? Full scoring criteria for how these platforms are evaluated live at /methodology.
Track sentiment before the recession finds you
A share of voice number going up cannot tell you whether the trust story above is already happening inside your own brand’s mentions. Only a sentiment layer can.
Pull your last quarter of AI-generated brand mentions and score them for tone, not just count. Compare that tone against the trust numbers in this piece. If your sentiment is flat while consumer trust in the channel is falling 28 points, you are due for a closer look.
The adoption curve is not going to reverse. Usage is climbing, Pew’s 49% adoption figure will keep growing, and AI search is not going away as a research channel. Trust is the number that is actually moving, and it is the number your current dashboard was probably never built to catch.
Compare how the leading AI visibility platforms handle sentiment and perception tracking, not just raw mentions, at /rankings/ai-visibility-tools.