Investigative Report Reveals Over 4,000 AI-Generated Fake News Websites Manipulating Google Algorithms for Ad Revenue
French investigative journalist Jean-Marc Manach uncovered the existence of more than 4,000 fake news websites created using generative artificial intelligence techniques. These sites aim to manipulate Google’s search algorithms, particularly Google Discover, to generate substantial advertising revenue.
According to Manach, these sites are widespread across France and include over 100 English-language websites. Their content consists of either plagiarized articles from other sources or entirely fabricated stories. Behind these operations are SEO (search engine optimization) experts seeking profits through backlink schemes and visibility on Google Discover, a news feed tailored for Google app users on Android devices.
Manach began his investigation in early 2024 after noticing a surge of unfamiliar websites. Collaborating with his journalism students, he identified 250 sites by October, and later expanded the probe with the French newspaper Libération, ultimately uncovering over 4,000 such websites. He believes at least two site owners have become millionaires from advertising revenue, despite the content lacking any real editorial oversight.
Manach points out that Google Discover, which recommends content based on user interests, fails to distinguish fake sites from legitimate ones, allowing these sites to appear as credible sources. Some editors refer to Discover as a “cash machine,” capable of generating thousands of dollars daily from ads. Reports indicate Discover has become a primary traffic source in the UK, surpassing traditional search referrals.
Among the fabricated news promoted by these sites are false claims such as the abolition of cash in France, banning grandparents from transferring money to grandchildren, and the French government withdrawing savings to fund the Ukraine war. Other bizarre stories include the discovery of a 25,000-year-old pyramid beneath a mountain and scientists finding a giant predatory bird-like creature frozen under Antarctic ice.
These articles often feature AI-generated images and sensational headlines designed to provoke curiosity or fear, increasing click-through rates.
Manach attributes the rise of this phenomenon partly to YouTube tutorials from 2023 and 2024 teaching how to “hack” Discover using generative AI. He warns that the French market is now “saturated” with such sites, pushing some operators to target foreign markets in other languages.
He has identified over 120 companies and editors behind these websites, including journalists and media trainers. Some sites previously employed real journalists but replaced them with AI-generated content without informing their audiences.
In response to these allegations, Google stated it enforces strict policies against unwanted content, whether human- or AI-produced, and its systems work to reduce low-quality content visibility on Discover and search results.
However, Manach argues these measures are insufficient. He stresses the need for journalists to demonstrate the distinctiveness of their work from AI content and for audiences to be informed when content is artificially generated.
Concluding his investigation, Manach calls for clear labeling of AI-produced content and penalties for those who conceal it. He warns that this issue threatens not only journalism but also democracy and the public’s right to reliable information.
He hopes that more journalists and fact-checkers worldwide will document these sites and combat the growing “information pollution” spreading rapidly across digital spaces.
