The Reuters Institute’s Digital News Report 2025 lands at a crucial moment for news media. Based on nearly 100,000 survey respondents across 48 markets, this year’s report is the most extensive to date, covering regions that together account for more than half the world’s population. Its breadth allows for a fascinating picture of global news consumption trends, but the standout theme for anyone interested in the impact of artificial intelligence (AI) is the report’s first systematic exploration of how AI is beginning to reshape the production, distribution, and consumption of news.
AI is not a footnote in the broader story of platform disruption, declining trust, or business model strain. It is central to questions of audience engagement, publisher viability, and journalistic integrity. The report shows both the opportunities and the risk of this shift – a world in which news organisations are exploring AI to personalise content, summarise complex stories, and translate material for multilingual audiences, even as readers voice scepticism about trust and accuracy in AI-generated news.
At the same time, AI is emerging as an independent distribution channel, not just a tool for newsroom production. For the first time, the report specifically asked about AI chatbots as a news source. The results are striking: while only 7% of respondents across all countries report using AI chatbots for news on a weekly basis, this figure jumps to 15% among the under-25s. This generational shift emphasises an existing dynamic. Younger users, already the least loyal to legacy media brands, are adopting AI-powered conversational interfaces at significantly higher rates. The trend threatens to further fracture the direct relationships publishers have tried to maintain with their audiences, in the same way that earlier shifts to social media and search weakened those bonds.
But the uptake of AI news interfaces is only part of the picture. The report also explores the complex attitudes that audiences have toward AI in the news and journalism more widely. Survey respondents broadly expect AI to make the news cheaper to produce – a finding with a net positive difference of 29 percentage points. There is also a net expectation that AI will make news more up-to-date (+16). But these perceived benefits are accompanied by clear concerns about the integrity of the news product itself. Respondents anticipate that AI will make news less transparent (-8), less accurate (-8), and less trustworthy (-18).
The report shows that these fears are pervasive, with qualitative research underscoring that many audiences remain much more comfortable with AI use cases where humans remain in the loop. For example, there is relatively strong audience enthusiasm for AI applications like summarising the news (27%), translating stories into different languages (24%), offering better story recommendations (21%), and enabling chatbots to answer questions about news (18%). These use cases are perceived as enhancing accessibility and relevance rather than replacing core journalistic judgement.
But despite this apparent openness to assistive AI features, publishers do face a dilemma. The same tech advances that enable them to provide improved personalisation risk undermining the basis of traditional news consumption. Some survey respondents worry that heavy reliance on AI-driven recommendations will mean missing out on important stories, pointing to a tension at the heart of AI adoption in journalism: the need to balance efficiency and personalisation with editorial breadth and public interest goals.
Beyond the question of content production, the Digital News Report 2025 highlights another element of AI’s impact: distribution and audience capture. The rise of generative AI features in search engines and other platforms represents another threat to publisher control over traffic. As large companies integrate real-time news summaries generated by AI directly into their offerings, publishers risk losing referral traffic to their own websites and apps. While the report does not provide precise traffic loss figures, it documents publishers’ anxieties about being further disintermediated.
This concern is all the more marked in a media ecosystem already dealing with declining engagement with traditional channels. The report notes that use of TV, print, and news websites continues to fall in most markets, while dependence on social media, video platforms, and online aggregators is rising. Indeed, the data show that globally, around a third of the sample use Facebook (36%) and YouTube (30%) for news each week, with TikTok use for news growing especially fast, reaching 17% overall but almost half the online sample in Thailand. Social media and video networks are also the primary news source for 44% of 18–24-year-olds, compared with just 20% for those over 55.
AI is both implicated in and accelerated by these distributional shifts. The major platforms are increasingly integrating AI-powered features – short-form video recommendations, automated news summaries, and conversational search assistants – that further disrupt the relationship between audiences and publishers. The Reuters Institute’s survey shows how this compounds the attention crisis for traditional outlets, who find it ever harder to maintain direct loyalty as users engage with news in fragmented, platform-centric ways.
Compounding the challenge, the report details how audiences are increasingly aware of – and concerned about – the problem of false or misleading information online. Over half of respondents (58%) say they remain worried about their ability to tell what is true from what is false, a proportion that rises to 73% in Africa and the United States but is still 46% even in Western Europe. When asked about the underlying sources of false or misleading information, 47% of respondents worldwide identify online influencers and personalities as the biggest threat, tied with national politicians.
It is within this climate of misinformation and fractured trust that AI-based news interfaces are emerging. The report notes that while younger people are more willing to experiment with using AI chatbots to check information, trusted news brands and official sources remain the most frequently cited destinations when people want to verify whether something is true or false online. This suggests a complicated dynamic: audiences may be willing to adopt AI interfaces for convenience, but they remain wary about the quality and provenance of the information those interfaces provide.
Publishers, too, are aware of the tensions. Many are hoping that AI might actually increase the perceived value of human-generated journalism by drawinga clearer contrasts with synthetic content of uncertain origin. The report suggests that while AI might make some forms of news cheaper and faster to produce, the reputational premium for trusted brands could grow in an environment awash with machine-generated text. However, this is far from guaranteed. The same report documents publishers’ struggles to grow digital subscription businesses, with the proportion paying for any online news stable at 18% across a basket of 20 richer countries. Even in Norway, the global leader, only 42% pay.
One of the starkest takeaways in the Digital News Report 2025 is that younger audiences are not only fragmenting across platforms but also diversifying their modes of consumption. Video and audio formats are gaining ground. The proportion consuming social video news weekly has risen from 52% in 2020 to 65% in 2025, while consumption of any online video news has grown from 67% to 75% over the same period. In the United States, weekly news video consumption jumped from 55% in 2021 to 72% today. These shifts are also being accelerated by platform strategies that now prioritise video, with algorithms on Facebook, Instagram, and X pushing visual content more heavily, while Google has added a short video tab to search results.
AI is not separate from these changes; it is the enabler. Recommendation systems, summarisation engines, automated transcription and translation – all these AI-driven processes are making video content more discoverable, more personal, and arguably more addictive. The result is a further weakening of the already strained link between publishers and their audiences. As platforms perfect AI-powered curation and interaction, the brand value of individual news outlets risks being subsumed within a sea of generic or repackaged content.
Meanwhile, even as audiences express anxiety about misinformation, they often pay more attention to creators and influencers than to traditional media brands on the very networks they use for news. The report highlights that on Facebook, 44% of users pay most attention to traditional news media when consuming news on the platform, but that proportion falls to 25% on Instagram and only 40% on TikTok. In Kenya, for example, creators and influencers play a much larger role in news consumption on TikTok than they do in Norway, where traditional brands still command more attention.
This uneven global picture complicates any simple narrative about AI and news. In some markets, public service media and strong legacy brands have invested in adapting AI-powered and video strategies to maintain relevance, with mixed success. For example, Norwegian brands such as VG have used TikTok to repackage complex investigative stories into visually engaging formats designed for Gen Z. By contrast, in many markets in the Global South, barriers to literacy and data cost have encouraged direct consumption of video and chat interfaces, leapfrogging traditional text-based models entirely.
The report also contextualises AI within a wider battle over platform strategy and control. As social media companies such as Meta oscillate between investing in news and retreating from it in favour of creator content, publishers have had to adapt to successive waves of algorithmic change. The integration of AI-driven features is only the latest in a series of shocks that reduce publisher leverage over distribution. And with generative AI accelerating the commodification of news summaries, the challenge of sustaining differentiated, valued journalistic brands becomes even steeper.
Finally, while the Digital News Report 2025 notes that survey-based results are self-reported and subject to biases, the sheer scale of the data – covering six continents and almost 100,000 respondents – gives real weight to its conclusions. The rise of AI in news consumption isn’t hypothetical, it’s here now, and its impacts are already measurable. The appearance of AI chatbots as a meaningful source of news, especially among the under-25s, is only one clear signal of a broader shift.
The news industry’s relationship with AI is defined by complexity and contradiction. On the one hand, AI offers ways to make journalism cheaper, more efficient, and more personalised – capable of meeting the demands of audiences accustomed to instantaneous, tailored content. On the other, it threatens to erode the trust, transparency, and editorial independence on which journalism depends. The Digital News Report 2025 does not offer easy answers. Instead, it sets out a rich, evidence-based portrait of an industry grappling with profound technological change at a moment when the stakes for democracy and public debate have rarely been higher.