AI-personalized news takes new forms (but do readers want it?)

Many newsrooms already use generative AI for efficiency and back-end tasks. Now they’re increasingly setting their sights on using AI to help deliver news that is more personally relevant and accessible for audiences, at a time when news interest has waned and avoidance has arisen in many countries.

This is not a new phenomenon. Many outlets have been personalizing news recommendations for years, and while AI can help enhance tools for tailoring news selection, the more substantive shift introduced by generative AI is the possibility to personalize news formats.

Some newsrooms have integrated AI tools into their websites that allow audiences to automatically summarize news articles (e.g. Aftonbladet in Sweden) or change news text to audio (e.g. the Miami Herald in the U.S.). Argentinian newspaper Clarín uses a tool called UalterAI to offer a variety of additional analyses, including key quotes and figures, as well as a glossary for technical terms. Beyond personalized selection and formats, newspapers like The Washington Post have been trialing tools that can answer complex queries based on their own archives.

These experiments and others are likely to continue. According to the Reuters Institute’s recent survey of newsroom leaders, published in this year’s report on media trends, four out of five respondents anticipated using AI to improve recommendations in the upcoming year. Most of them said they would be actively exploring the use of AI for audience-facing products that would allow changing text to audio (75%), summarization (70%), translation into different languages (65%), and use of AI chatbots (56%).

However, the success of these tools will largely depend on how keen audiences are to use them in the first place — something we looked into in our Digital News Report 2025, which was published earlier this month.

We asked survey respondents across 48 countries about their interest in eight different kinds of AI-powered personalization. Our aggregate results show interest is relatively low (below 30%) for any single option, but over half in most countries were interested in at least one.

Interest was highest in options geared toward making news consumption more efficient or relevant for audiences: Summaries (27%) and translations (24%) were at the top of the list, followed by customized news homepages and story recommendations and news alerts (21% each). Audience interest was lowest in the format conversion options such as text-to-audio — which, as noted above, was at the top of the list of initiatives news industry leaders planned for 2025.

Newsrooms are juggling a number of considerations when choosing what to prioritize, and features like the text-to-audio option may be seen as cheaper and less risky to implement than others. It’s worth keeping in mind, though, that 19% of our respondents said they were interested in none of the options and 15% said they didn’t know.

Regional and demographic differences

When we zoom in and look at results by country, we see some interesting differences. First, we found that interest in AI personalization was generally lower in places where people tend to be more uncomfortable with using news made by AI. One way to look at this is by plotting the proportion of people who are uninterested in AI personalization (the percentage of people who said they are interested in none of the options provided) against the proportion of those who are uncomfortable using news made mostly by AI, at the country level. The strong correlation between the two suggests that interest in AI personalization may be shaped in part by broader attitudes toward use of AI in news.

For example, in the U.K., the country in our sample with the largest proportion of people who are uncomfortable using news made mostly by AI (64%), the percentage of people interested in none of the AI personalization options is the highest at 41% — more than twice the global average. We see the opposite in places like India and Thailand, where fewer people are uncomfortable with AI and more people mention at least one option.

There are also cross-national differences in the relative popularity of AI personalization options, reflecting distinct needs or preferences within markets. While summarization featured at the top in most places, translation tended to rank higher in linguistically unique European countries with smaller populations, such as Finland and Hungary. In countries like the U.S., translation ranked below the global average.

Meanwhile, the option for adapting the language of news for different reading levels was higher in countries with lower literacy or reading levels. In countries like the Philippines, Kenya, and Nigeria it was among the top three, and in India it was the most popular option, selected by over one-third of respondents. It ranked in the bottom three in Japan and the U.S. These differences highlight distinct needs that may stand out more in some countries than others.

Country differences aside, we also found variation across age groups. Whereas 66% of respondents globally selected at least one AI personalization option, this percentage rose to 76% among people under 35. These age differences were most prominent when it came to the use of AI for personalizing news formats and AI chatbots, and may be driven at least in part by familiarity and comfort with AI, which tends to be higher among younger people. This suggests news organizations may want to focus such kinds of initiatives on younger audiences.

Could AI personalization help encourage news use among those most disengaged? Our data suggest that interest across the board tends to be lower among those least interested in news and those who avoid news more frequently. That said, particular forms of personalization may be welcome by small subsets of news avoiders, such as those who find news difficult to understand. These respondents were almost twice as likely to select the option for modifying language for different reading levels compared to the overall average (32% vs 17%).

Overall, our findings suggest we might expect an uneven uptake of AI-driven personalization initiatives. News organizations may want to direct particular types of tools and products toward younger audiences, who seem more receptive to them. Likewise, things may move far more quickly in countries where newsrooms and audiences have been less conservative about the use of AI in journalism, such as India and Thailand, compared to markets like the U.K. and Germany, where audiences are much more skeptical.

We are likely to see some differences in the types of AI personalization options that resonate most with audiences across different national contexts depending on particular local needs and preferences. However, given the fairly low interest in any single option we asked about, newsrooms may have more success by providing a selection of AI-personalization options users can choose from.

Just as expressed interest in a tool doesn’t always translate into actual use, a lack of interest doesn’t necessarily mean people won’t try something. It’s also possible that low interest in some of these options can be explained by a lack of understanding of what they would entail or look like in practice.

However, the noticeable gap between the top AI initiative planned by newsroom managers and the most popular options among audiences suggests newsrooms need to keep a finger on the pulse of what kinds of initiatives will create value or solve problems for their audiences in practice.

You can read the full analysis in the Reuters Institute Digital News Report 2025.