Google has promised it is working on a “fix” for spam sites in its Discover app after more fake news stories topped the rankings in the UK last week.
Fake stories including “Free TV Licence for UK Over-60s – How to Apply Under New Rules”, from a site called “mountainbrooksolarproject”, was one example at the top of the Discover rankings, visible to 2.95 million people.
AI copy detector Pangram assessed the story as being “fully AI generated”. It is also false, with no plans to introduce free TV licences for over-60s.
Google earns advertising money from the Discover feed, as do publishers of stories, which has made it an attractive target for spammers making up fake news stories to earn ad money.
Google Discover recommends content to users on Android devices and in Google apps, based on their searches and activities in other apps, and works like an automated news feed, without human oversight.
A similar version of the same story, hosted on a different site, was the top story over the weekend, while another false story saying “UK Bus Pass Rules Change Started From 5th November 2025 – Full Details” was viewed by 1.85 million people.
Searching for “UK Bus Pass Changes November 2025” brings up dozens of sites hosting the same false story, presumably with intent to rank in Google Discover.
The stories are hosted on sites ranging from events sites – Gone Coastal Events – to lapsed projects – Infinite Transitions Project to schools – Prashant High School.
A Google spokesperson said: “We keep the vast majority of spam out of Discover through robust spam-fighting systems and clear policies against new and emerging forms of low quality, manipulative content.
“We’re actively working on a fix that will better address the specific type of spam that’s being referenced here, maintaining our high bar for quality in Discover.”
Malcolm Coles, a British consultant who helps deliver audience growth for publishers, spotted the fake stories ranking highly for visibility (a measure of how many people can see the stories in their Android phone apps) using software tools Marfeel and Newzdash.
Coles told Press Gazette: “They’re probably buying lapsed domains and reinvigorating them – and then using deliberately alarming or surprising headlines to get people to click, followed by giving information so detailed that it feels like it must be reliable.
“But it’s all untrue – the state pension age hasn’t changed and over 60s can’t get free TV licences.”
Coles believes that Google needs to adapt its processes to prevent spammers buying lapsed domains to rank highly in Discover.
He said: “Google needs to either go back to only showing trusted publisher domains in Discover or else put much more robust filters in place around factual accuracy and trust. It’s ridiculous that newly (re-)registered domains are allowed into Discover like this.
For publishers in the UK, the Discover feed has become an increasingly important source of traffic as traffic from Google search has declined, with Reach describing Discover as the “biggest referrer” of traffic in late 2024. More than two-thirds of Google traffic to the biggest news websites globally is now coming from its Discover feed, Chartbeat data showed in August.
What sort of sites can rank in Discover, according to Google?
Publishers often think of Discover as something akin to a news aggregator, but a Google spokesperson told Press Gazette that it shows a mix of content based on user’s interests.
This can include anything from social media posts to short videos and more, and is not restricted to content from news publishers or publishers in the conventional sense at all.
Some publishers, such as the music nostalgia-focused Far Out magazine, enjoy great success on Discover’s feed, but Google will display any content that is indexed by Google and meets its standards.
These standards bar hateful and explicit content and advertising that is not sufficiently marked – as well as misleading content. They do not specifically mention AI-generated content.
Recent updates to the feed have prioritised more creator content and social media posts over publisher content.
In a recent blog post, Google said: “ In the coming weeks, you’ll start to see more types of content in Discover from publishers and creators across the web, such as posts from X and Instagram and YouTube Shorts, with more platforms to come.
“In our research, people told us they enjoyed seeing a mix of content in Discover, including videos and social posts, in addition to articles.”
Google told Press Gazette it regularly updates spam policies and reviews sites that are flagged and take action if they break policies.
How are spammers ‘hacking’ Google Discover?
Coles told Press Gazette he believes spammers are buying lapsed domains and reinvigorating them as a platform to launch spammy stories into Discover.
Jean-Marc Manach, a French data journalist who is tracking an explosion in AI-generated news sites in the country, said spammers have “several tactics” to rank in Google Discover.
Manach’s research has uncovered 8,300 AI-generated sites in France, many of which attempt to use Google Discover as a “cash machine” to rake in advertising revenue.
Mamach said: “There are several tactics: most of them buy expired domain names which used to be indexed in Google News (or, better, Discover).
“They install a WordPress Lite theme optimised to Google’s specifications in order to speed their appearance on mobile phones (as Discover is only available on mobile phones, for now).
“They analyse what’s trending on social networks, generate several articles about those trends, illustrated by large (1,200 pixels) images with clickbait titles, and hope that one (or several) of them will ‘pop’ on Discover.”
Manach said the spammers rely on what they call the “spark” (“étincelle” in french), where they manually generate clicks to fool Google into thinking stories are popular.
He said: “Discover’s algorithm relies, among other characteristics, on the fact that people click on links.
“In order to let Google think that lots of people click on those, some editors post their links on Facebook groups and/or buy fake traffic on farms dedicated to artificially clicking on links.”
Mamach’s database of fake sites, which spew plagiarised AI-generated stories and fake news, now includes more than 8,300 in French, 300 in English and 150 in German.
Email pged@pressgazette.co.uk to point out mistakes, provide story tips or send in a letter for publication on our "Letters Page" blog
