A Morning Consult report published May 10, 2026 found that 63% of Americans trust AI only a little or not at all, and marketing circles have been using that number ever since to argue against investing in AI search visibility. The argument is tidy. It is also built on roughly one-third of what the report actually says.
The study itself is serious: a nationally representative sample of 1,048 U.S. adults, daily data collection, continuous brand tracking across more than 3,000 brands in 40-plus countries. This is not a LinkedIn poll. The headline number is real and it deserves to be taken seriously. What it does not deserve is to be treated as a complete picture when it is closer to a chapter heading.
Key Facts
- Published: May 10, 2026
- Sample: 1,048 U.S. adults, nationally representative
- Scope: 3,000+ brands tracked across 40+ countries
- Headline finding: 63% of Americans trust AI only a little or not at all; "not at all" jumped 9 points to 36% since August 2025
- Usage trend: Self-reported chatbot use rose from 31% (May 2024) to 49% (April 2026)
- Trust peak: Net trust score of 12 in October 2025, softened to 8.5 by April 2026
- Source: Should You Actually Care About AI Search?
The share of Americans saying they do not trust AI at all jumped 9 points to 36% since August 2025. AI ranks 10th out of 198 tracked categories in awareness-adjusted distrust, sitting alongside tobacco, chemicals, dating apps, and crypto. Read in isolation, that is a damning portrait. But the report tracks three distinct types of trust, and they point in three different directions. Most people sharing the headline have only read one of them.
Brand trust is falling. Seven of the ten leading AI brands lost net trust year-over-year. xAI fell 5.5 points. Perplexity dropped 5.4. Meta AI lost 5. OpenAI slid 2.2. Only Gemini (+6), Anthropic (+1.4), and Apple Intelligence (+1.2) gained ground. Twelve percent of Americans say AI leaders are generally credible and trustworthy. This part of the story is real, and it is getting worse.
But industry trust is tangled up in something else entirely. The top negative associations driving distrust are: spreading misinformation or fake content (39%), threatening jobs (38%), collecting and misusing personal data (33%), and training AI on data without consent (32%). These are concerns about what AI does to society, not assessments of whether it helped someone research a vendor or summarize a document. People are answering a cultural question when they say they distrust AI. The report captures both the cultural and the utilitarian in the same number, and almost nobody sharing it has noticed.
Utility trust tells the opposite story. Self-reported chatbot use jumped from 31% in May 2024 to 49% in April 2026. Nearly half of American adults are using AI tools with some regularity. Expected AI impact registers net positive in healthcare (+14), daily life (+7), and education (+5). The most clarifying data point in the study: when asked who they trust to use AI responsibly, Americans gave net positive scores to exactly three entities — themselves (+16), their family (+7), and their friends (+5). AI companies sit at -14. Large tech lands at -18. The federal government is at -28. People do not trust AI companies. They trust themselves to use AI just fine. That distinction is carrying enormous weight that almost no one is acknowledging.
What makes this especially interesting is that trust and usage are now moving in opposite directions. Net trust peaked near 12 in October 2025 and has since softened to 8.5. In that same window, chatbot use climbed from 31% to 49%. The trust softening traces to a specific cluster of late-2025 headlines: AI-attributed layoffs, mental health and privacy incidents involving chatbots, the expanding reach of deepfakes. These are anxieties about AI's cultural footprint. They are not changing whether people open ChatGPT to research a purchase. The customer who tells a pollster they distrust AI companies is often the same person using one to compare your product against a competitor's. Both things are true simultaneously, and the data shows it plainly.
Subjective
What bothers me about how this report circulates is that it gives people permission to feel rigorous while making an uninformed decision. Citing a real number to justify a predetermined conclusion is not analysis. It is dressed-up inertia. The self-trust finding is the one I keep returning to: +16 for themselves, -14 for AI companies. Your customer does not trust OpenAI. They trust their own judgment when they use it. That is not a reason to ignore AI search. It is actually a reason to make sure you show up when they go looking, because they are already going. They just believe they got there on their own.