The State of AI Search in 2026: Statistics and Trends

Someone inyour target audience asks ChatGPT to compare two specific products before buying, and acts on the recommendation without ever opening Google. Somewhere else, a person reads the AI summary at the top of a Google results page, gets their answer, and closes the tab. Neither of them touched a search result the way people did for the last twenty years. That’s becoming the norm, and it’s easy to see why AI search has become the most consequential shift the industry has faced in a decade.

The scale is hard to grasp. ChatGPT reaches almost a billion users a week, Google’s Gemini has caught up faster than almost anyone predicted, and AI features inside Google Search answer questions for over a billion people a month. But the size of the audience isn’t the only interesting part. What matters is how this audience behaves once it gets there, which sources these tools trust, and who wins or loses when an answer shows up before a single link does.

We pulled this report together from OpenAI and Google company disclosures, Pew Research Center, clickstream data from Datos and SparkToro, retail analytics from Adobe, and a survey of more than 2,000 US adults run in April 2026. What follows is the clearest picture we could assemble of where AI search actually stands in 2026 and what it means if you’re trying to get found.

Key Stats at a Glance

• ChatGPT and Google’s Geminin app each reach more than 900 million weekly active users
• 31% of Americans now interact with AI at least several times a day
• ChatGPT holds about 55% of worldwide AI chatbot traffic
• About 68% of Americans trust AI search results as much as or more than Google, and 79% have made a real decision based on what an AI tool told them
• Only 18% of AI users say they consistently verify what AI tells them, and fewer than half check the sources it provides
• The number of Google searches per US user fell nearly 20% from 2024 to 2025
• AI referral traffic to US retail sites jumped 693% year over year during the 2025 holiday season.
• Leading AI assistants misrepresented news content in 45% of responses in a study of 3,000 answers across 14 languages.

How Many People Use AI Search?

Adoption has moved fast, but not evenly. Pew Research Center found that 34% of US adults had ever used ChatGPT by early 2025, about double the share from 2023, including 58% of adults under 30. The broader picture is even larger once you count all AI tools rather than ChatGPT specifically. Pew also reported that 31% of Americans now interact with AI at least several times a day, up from 22% in early 2024.

Younger users are pulling the average up. Roughly two-thirds of teens ages 13 to 17 say they use an AI chatbot, with information seeking and schoolwork the most common reasons.

A Clear Spark Digital survey of 2,127 US adults found that 48% use AI tools daily and 76% use them at least weekly. Daily use was high across every age group, and one counterintuitive finding stood out. Adults 61 and older reported the highest daily usage rate of any group, ahead of Gen Z.

(Sources: Pew Research Center, Pew Research Center, Pew Research Center, Clear Spark Digital)

Platform Growth

The most popular AI platforms have grown incredibly quickly, but not every company shares exact details and figures. The clearest details belongs to ChatGPT, because OpenAI publishes its numbers publicly. The ground-breaking chatbot reached 700 million weekly active users by July 2025 and crossed 900 million weekly active users by late February 2026. OpenAI also reported 50 million paying consumer subscribers at that point.

Google’s Gemini has closed the gap quickly. The Gemini app surpassed 750 million monthly active users by the end of 2025, up from 650 million the prior quarter. By Google I/O in May 2026, the app had reached 900 million monthly active users, and AI Mode inside Google Search passed one billion monthly users on its own. That search number probably matters more than the app figure, because it shows how much of Google’s enormous search audience is now interacting with AI answers directly.

Perplexity is harder to pin down because the company doesn’t publicly disclose its numbers. The figures available come mostly from public statements relayed through reporting, which point to about 780 million monthly queries by May 2025. Treat any Perplexity user count you see as an rough estimate rather than a confirmed number.

(Sources: NBER, TechCrunch, TechCrunch, StartupHub.ai, theStacc, Index.dev)

AI Search Market Share

Market share estimates vary widely depending on how activity is measured, so the numbers below all come from Similarweb’s web-visit data for consistency. They should be taken as estimates that give a rough idea, rather than exact statistics.

By web visits, ChatGPT held about 54.7% of worldwide AI chatbot traffic in April 2026, down from roughly 76.5% in February 2025. Gemini sat second at 27.4%, up from about 5.6% in February 2025. The United States skews more toward ChatGPT, where it held 58.9% against Gemini’s 19.2% and Claude’s 12.5%.

The rest of the field is smaller but worth knowing. Worldwide, DeepSeek held 4.1%, Grok 2.8%, Perplexity 1.5%, and the standalone Copilot web app 1.3%. Claude was the fastest-growing major chatbot by web visits, up about 306% in a single quarter off a small base.

(Sources: Momentic, Similarweb)

How AI Is Changing Search Behavior

The Datos and SparkToro State of Search reports, built on the browsing behavior of tens of millions of real users, found that AI tool visits grew almost 50% year over year and reached 0.77% of all US desktop activity by December 2025. Traditional search visits grew too, across Google, Bing, Yahoo, and DuckDuckGo. They conclude that AI is additive rather than substitutive. People aren’t trading Google for ChatGPT. They’re using both for different jobs.

That said, behavior is shifting. The same research found that the number of Google searches per US user fell nearly 20% from 2024 to 2025, though the drop was only 2 to 3% in the EU and UK. The change is that AI answers, including Google’s own AI Overviews, resolve a question before the user needs to click a result or run a second and third search.

This finding is supported by looking at what users are turning to AI for. OpenAI’s research, based on 1.1 million conversations, found that practical guidance, seeking information, and writing account for nearly 80% of ChatGPT conversations, and that non-work use rose from 53% to 73% over the study period. A large share of AI use is exactly the kind of information seeking people used to type into a search box.

(Sources: Datos, Search Engine Land, NBER,

Trust and Verification

People trust AI answers more than the technology’s track record justifies, and they check those answers far less than they should, based on a study by Clear Spark Digital.

About 68% of Americans said they trust AI search results as much as or more than traditional Google results. That’s a remarkable level of confidence in a category that’s been mainstream for only a few short years. And it translates into users taking action, since 78.5% of respondents reported making a real-life decision based on something an AI tool told them.

The verification side is also interesting. Only 18% of AI users said they always verify AI-generated information, and fewer than half manually check the sources that AI tools provide. When people do verify, they lean on the old organic search. 72% turn to Google, and even among people who say they trust AI more than Google, 59% still use Google to verify.

(Sources: Clear Spark Digital)

What AI Search Tools Actually Cite

Several analytics firms have studied which sources AI tools cite, using large datasets. These studies come from companies that sell AI visibility tools, so there’s a commercial interest worth keeping in mind, but the patterns are consistent across providers.

A Profound analysis of 30 million citations across ChatGPT, Google AI Overviews, and Perplexity found that Wikipedia leads ChatGPT’s citations, while Reddit leads on AI Overviews and Perplexity. User-generated content and reference sites dominate, not the publications you might expect.

A separate audit by the firm 5W, synthesizing nine datasets, had similar findings. Wikipedia and Reddit together drive more than 25% of US ChatGPT citations, and the Wall Street Journal, New York Times, Bloomberg, and Financial Times don’t appear in the top 20 at all. Forbes is the only US business publication on the list, ranked 18th.

The audit also found two findings any business can act on. Listings across review sites like G2, Capterra, Trustpilot, and Yelp correlated with roughly a 3x citation lift, and YouTube correlated at 0.737 with AI visibility, the strongest single predictor in any recent study, based on an Ahrefs analysis of 75,000 brands.

One caution. These citation patterns are volatile. Semrush tracked Reddit’s ChatGPT citation share falling from close to 60% of prompt responses to around 10% in a matter of weeks in September 2025. What gets cited can shift fast, so treat any single snapshot as a moment in time.

(Sources: Profound, PR Newswire, Ahrefs, Semrush)

AI Referral Traffic, the Two-Sided Story

AI is sending traffic in two opposite directions at once, and which side you’re on depends entirely on what kind of site you run.

For retailers, the news is good and getting better. Adobe Analytics, which tracks more than a trillion visits to US retail sites, found that AI referral traffic to retail sites rose 693% year over year during the 2025 holiday season, with travel up 539% and financial services up 266%. That traffic also performs. AI referrals converted 42% better than non-AI traffic in March 2026, a sharp reversal from a year earlier when AI traffic converted 38% worse. People arriving from an AI tool have done their research already.

For publishers, the same technology negatively impacts traffic. Pew Research analyzed the browsing data of 900 US adults and found that about 18% of Google searches produced an AI summary in March 2025. When a summary appeared, users clicked a traditional result in just 8% of visits, compared with 15% when no summary was present, and only 1% clicked a link inside the summary itself. Worth noting that Google disputed the study’s methodology, calling the query set unrepresentative.

(Sources: Adobe, Adobe, Pew Research Center, PPC Land)

How Accurate Is AI Search?

Given how much people trust and act on AI answers, the accuracy deserves a hard look. So far, it’s not reassuring.

The largest study of its kind, run by the European Broadcasting Union and the BBC, had professional journalists assess 3,000 AI responses across 14 languages. 45% of responses contained at least one significant issue, 81% had some kind of problem, and 20% had major accuracy issues including hallucinated details and outdated information. Gemini performed worst, with significant problems in 76% of its responses, largely from poor sourcing. One example the researchers cited was ChatGPT naming Pope Francis as the current Pope months after his death.

This matters more as AI becomes a primary news source. The Reuters Institute found that 7% of online news consumers now rely on AI assistants, rising to 15% among people under 25. Consider this along with the trust and verification numbers from the Clear Spark Digital survey mentioned above. Most people trust these tools and rarely verify them, while the tools get something significant wrong close to half the time when it comes to news.

(Sources: EBU, AOL,)

What’s Next?

The pace is uncertain, but the direction is obvious. Semrush projects that AI search visitors will overtake traditional search visitors by 2028. This is just a forecast, not a measured trend, but it also shows how massive the shift is.

The data is clear that AI is changing how people find and trust information faster than most businesses are adjusting to it. The companies that benefit will be the ones that understand where their audience actually looks for answers, what those tools cite, and how to be present in both the AI answer and the traditional result.

(Source: Marketing Charts)