[ad_1]
AI answers from OpenAI, Google and Perplexity draw on a “narrow range” of the biggest publishers when responding to news queries, according to new research from UK thinktank IPPR.
Looking at Google Gemini, Google AI Overviews, ChatGPT and Perplexity, the research found that on average 34% of journalistic citations on each tool go to only one newsbrand.
On average the most-cited news source on each platform was four times more prominent than the next highest outlet by source links.
The IPPR “AI’s Got News For You” report said: “AI creates new winners and losers, with each AI tool prioritising news brands in different ways, in each case foregrounding a distinct selection of news outlets compared with those that are currently most popular across the UK.”
window.addEventListener(“message”,function(a){if(void 0!==a.data[“datawrapper-height”]){var e=document.querySelectorAll(“iframe”);for(var t in a.data[“datawrapper-height”])for(var r,i=0;r=e[i];i++)if(r.contentWindow===a.source){var d=a.data[“datawrapper-height”][t]+”px”;r.style.height=d}}});
On Google’s AI Overviews some 41% of news links cited the BBC, its top source.
The BBC was also the most-cited news source on Perplexity, where it made up 31% of publisher links.
This is despite the fact the BBC blocks AI crawlers from Perplexity and other LLMs using the robots.txt protocol. The BBC threatened to sue Perplexity in June last year over the use of its content without permission.
Internet infrastructure provider Cloudflare accused Perplexity of getting around robots.txt blocks by using “stealth, undeclared crawlers”. Perplexity has insisted it “respects robots.txt directives” but that it partners with third-party crawlers and said it has “updated our agreements to ensure these providers also respect robots.txt, particularly for news publisher sites”.
Publishers currently cannot opt out of their content being used in Google’s AI Overviews without affecting their visibility in search. The Competition and Markets Authority said this week publishers should be able to do so and Google responded that it is “exploring updates to our controls to let sites specifically opt out of search generative AI features”.
However, publishers can opt out of Google’s Gemini tool, where the BBC did not appear at all in the study, without it affecting Google Search. The BBC was also not used as a source in OpenAI’s ChatGPT.
The most-cited source by both Gemini and ChatGPT was The Guardian, which made up 37% and 27% of publisher citations respectively.
IPPR analysed more than 2,500 links that appeared in response to 100 randomly generated news queries on four AI tools in October and November 2025. In Google’s case, AI Overviews were generated for 61 out of the 100 queries.
The sources analysed were those that were featured via in-text citations, rather than links at the bottom of an AI answer that may not have fed into the answer itself. A link was counted once if it appeared multiple times.
The news queries, which spanned topics from politics and current affairs to celebrities, entertainment and sport, were randomly generated by another AI model, Anthropic’s Claude Sonnet 4.5.
window.addEventListener(“message”,function(a){if(void 0!==a.data[“datawrapper-height”]){var e=document.querySelectorAll(“iframe”);for(var t in a.data[“datawrapper-height”])for(var r,i=0;r=e[i];i++)if(r.contentWindow===a.source){var d=a.data[“datawrapper-height”][t]+”px”;r.style.height=d}}});
window.addEventListener(“message”,function(a){if(void 0!==a.data[“datawrapper-height”]){var e=document.querySelectorAll(“iframe”);for(var t in a.data[“datawrapper-height”])for(var r,i=0;r=e[i];i++)if(r.contentWindow===a.source){var d=a.data[“datawrapper-height”][t]+”px”;r.style.height=d}}});
window.addEventListener(“message”,function(a){if(void 0!==a.data[“datawrapper-height”]){var e=document.querySelectorAll(“iframe”);for(var t in a.data[“datawrapper-height”])for(var r,i=0;r=e[i];i++)if(r.contentWindow===a.source){var d=a.data[“datawrapper-height”][t]+”px”;r.style.height=d}}});
window.addEventListener(“message”,function(a){if(void 0!==a.data[“datawrapper-height”]){var e=document.querySelectorAll(“iframe”);for(var t in a.data[“datawrapper-height”])for(var r,i=0;r=e[i];i++)if(r.contentWindow===a.source){var d=a.data[“datawrapper-height”][t]+”px”;r.style.height=d}}});
The report noted that the sources cited by AI tools “can change overnight”.
The Guardian has a deal with OpenAI giving it compensation for the use of its journalism on ChatGPT in short summaries and article extracts, with proper credit.
window.addEventListener(“message”,function(a){if(void 0!==a.data[“datawrapper-height”]){var e=document.querySelectorAll(“iframe”);for(var t in a.data[“datawrapper-height”])for(var r,i=0;r=e[i];i++)if(r.contentWindow===a.source){var d=a.data[“datawrapper-height”][t]+”px”;r.style.height=d}}});
The IPPR said: “Questions need answering around how financial relationships between AI companies and news brands shape AI answers.
“If licensed publications appear more prominently in AI answers, there is a risk of locking out smaller and local news providers who are less likely to get AI deals.”
Journalism outlets were the most prominent sources in the AI tools’ responses (making up 40%) ahead of government websites (23%).
window.addEventListener(“message”,function(a){if(void 0!==a.data[“datawrapper-height”]){var e=document.querySelectorAll(“iframe”);for(var t in a.data[“datawrapper-height”])for(var r,i=0;r=e[i];i++)if(r.contentWindow===a.source){var d=a.data[“datawrapper-height”][t]+”px”;r.style.height=d}}});
The IPPR report called for “standardised ‘nutrition labels’ for AI news content” so users are told how sources are selected and given more active choice about the answers they receive.
It said there is a lack of transparency about how AI tools choose what outlets to cite and that this means users are shown a “narrow set of publications without their knowledge”.
It also called for a “licensing market that supports a thriving news ecosystem and provides AI companies with access to diverse sources” and urged the Competition and Markets Authority to impose conduct requirements to bring Google to the table with news publishers under the powers it has by designating the tech giant with “strategic market status”.
It also suggested the creation of an “independent news licensing taskforce” could help with collective licensing.
News Media Association chief executive Owen Meredith said: “As the report demonstrates, weakening UK copyright law would deprive publishers of reward and payment for the trusted journalism that enables AI to be accurate and up to date. The Government must end the uncertainty it has created, by ruling out any new text and data mining exception in their March reports.
“The CMA must also intervene swiftly to stop Google using its dominant position to force publishers to fuel its AI chatbots for free. Fair payment from the market leader is critical to a functioning licensing market and to preventing big tech incumbents from monopolising AI.”
The CMA said earlier this week, as it set out its proposed remedies for Google including that publishers should be able to opt out of AI Overviews, that it would “wait for 12 months to see the impact of our initial measures before deciding whether to take further action in ensuring they receive fair and reasonable terms for their content”.
This means the CMA is not currently pursuing payment terms for Google’s use of publisher content.
The post AI answers cite ‘narrow range’ of top newsbrands led by BBC and Guardian appeared first on Press Gazette.
[ad_2]
Source link
