Executive summary and key findings
Our survey in six countries (Argentina, Denmark, France, Japan, the UK, and the US) explored how people use generative AI in their everyday lives, what they think its impact will be on different areas of society, and what they think about its use in news and journalism specifically. It is a follow-up to the survey that we conducted in the same six countries in 2024 (Fletcher and Nielsen 2024). Based on the results of this survey, and the previous one, we find the following.
Findings on the awareness and use of generative AI
The public’s use of generative AI has increased substantially in the last year. The proportion who say they have ever used a standalone generative AI system such as ChatGPT jumped from 40% to 61%, and weekly usage nearly doubled from 18% to 34%. ChatGPT is the single most popular AI system with 22% weekly usage, though adoption varies significantly by age. Information-seeking has become the primary use-case for AI, more than doubling to 24% weekly usage and surpassing media creation, while specialised uses like news consumption remain limited at 6%. Trust is concentrated among major brands, with ChatGPT leading the field again, though most users remain occasional rather than regular adopters.
In more detail, we find:
Awareness has surged year-on-year. The share of people across countries who have heard of at least one of 13 AI tools rose from 78% (2024) to 90% (2025); only 10% say they have heard of none. ChatGPT remains the most well-known brand, with pronounced country variation for others.
Use is expanding rapidly – especially weekly use – though not uniformly. Across countries, the proportion of people who say they have ever used any AI system rose from 40% (2024) to 61% (2025); weekly use nearly doubled from 18% to 34%.
Again, ChatGPT dominates active use. On average, 22% report using ChatGPT in the last week, well ahead of other tools. Since May 2024, the core user base for every major system has roughly doubled. Yet most people are not regular users: for the four most popular systems, many use them only monthly or once or twice ever, and large shares have never used or not heard of them.
Use of generative AI skews younger. In the 18–24 category, 59% used any generative AI in the last week, as opposed to 20% among those aged 55 and over, although this age gap is driven mainly by ChatGPT. Age differences are smaller for Google’s Gemini, and minimal for Copilot, Meta AI, and Grok – likely because they’re embedded in widely used products.
Information-seeking is now the lead use-case across countries. Weekly use of AI for getting information more than doubled (from 11% to 24%), overtaking creating media (up 7 percentage points to 21%). Social interaction is nascent but notable (7% overall; 13% of 18–24s versus 4% of 55+).
While specific information-retrieval tasks are broadening, with answering factual questions rising from 6% to 11%, media creation remains niche except for images. Image generation rose from 5% to 9% weekly; video (3%) and audio (2%) were basically flat; coding use was also flat, suggesting early adopters were already on board in 2024.
Getting news via a generative AI system has doubled but remains a minority activity. Weekly use rose from 3% to 6%, driven mainly by users in Japan and Argentina; it is strongest in Argentina and the USA and among 18–24s (8%) compared to 55+ (5%), and higher for degree-holders. Among AI-for-news users, ‘latest news’ (54%) and help with summarising, evaluating, or rewriting are most common. Younger users lean more towards using AI to help them navigate the news: 48% of 18–24s used AI to make a story easier to understand compared to 27% of 55+ (a 21 percentage point gap).
Trust is concentrated in a few brands. On average, 29% say they trust ChatGPT, ahead of Gemini (18%), Copilot (12%), and Meta AI (12%); most other brands are trusted by fewer than 10%, mostly due to low awareness. In most countries ChatGPT is more trusted than distrusted, with the exception of the UK.
Findings on public views on AI-generated search answers
AI-generated search answers have become commonplace in the six countries studied. User engagement with these AI answers is mixed, as only one third say they consistently click through to source links while 28% rarely or never do so, with younger users more likely to engage further. Trust levels are moderate at 50% among those who encounter AI answers, with users valuing their speed and information aggregation capabilities – although trust becomes conditional in high-stakes areas like health and politics, where many verify answers through traditional sources.
In more detail, we find:
People regularly see AI-generated search answers. Across countries, 54% say they saw an AI-generated answer to one of their searches in the last week. Reported weekly encounters are highest in Argentina (70%), followed by the UK (64%) and the USA (61%), and lowest in France (29%), where Google’s AI overview feature had not been rolled out at the time of fieldwork.
Self-reported click-through behaviour is mixed. Among those who saw AI answers, about one third (33%) say they always or often click links in the overview, 37% say they do so sometimes, and 28% rarely or never click through. Younger people are more likely to say they click through, although it is important to acknowledge that self-reported behaviour may differ from actual behaviour for some respondents.
Among those who have encountered AI answers, 50% say they trust them, with a significant minority remaining neutral. While there is little difference by gender, younger adults show slightly more outright trust. Respondents emphasised their speed and convenience and the fact that AI aggregates vast amounts of information as reasons to trust them.
Trust in AI answers in search is conditional, especially in high-stakes domains. Many say they verify answers, especially for health or politics, treating AI as a first pass before checking non-AI sources.