Google Search Ranks AI Spam Above Original Reporting in News Results

For example, I searched for “competing visions google openai” and saw a TechCrunch article at the top of Google News. Below that were articles from The Atlantic and Bloomberg comparing the rival companies’ approaches to AI development. But then, the fourth article to appear for that search, nestled just below those more reputable websites, was another Syrus #Blog article that largely copied the TechCrunch article in the top position.

As 404 Media reported in January, AI-powered articles appeared multiple times for basic queries earlier this year in Google News results. Two months later, Google announced major algorithm changes and new anti-spam policies aimed at improving search results. And in late April, Google announced that major tweaks to remove junk from its search engine ranking system were complete. “On April 19, we finished rolling out these changes. You’ll now see 45% less low-quality, unoriginal content in search results, compared to the 40% improvement we expected for this work,” Elizabeth Tucker, Google’s director of product management, wrote in a blog post.

Despite the changes, spammy content created with the help of AI remains a recurring and widespread problem for Google News.

“This is a really common issue on Google right now, and it’s hard to answer exactly why it’s happening,” says Lily Ray, senior director of search engine optimization at Amsive, a marketing agency. “We’ve had clients tell us, ‘Hey, they took our article and reworked it with AI. It looks exactly like what we wrote in our original content, but it’s just kind of gibberish rewritten by AI.’”

At first glance, it was clear to me that some of the images in Syrus’ blogs were generated by AI based on the illustrations’ droopy eyes and other distorted physical features—telltale signs of an AI trying to represent the human body.

Was the text of our article rewritten using AI? I reached out to the person behind the blog to learn more about how they did it and received an email confirmation that an Italian marketing agency created the blog. They claim to have used an AI tool as part of the writing process. “Regarding your concerns about plagiarism, we can assure you that our content creation process involves AI tools that analyze and synthesize information from various sources while always respecting intellectual property,” someone using the name Daniele Syrus wrote in an email.

They cite as sufficient evidence the single hyperlink at the bottom of the stolen article. While this is better than nothing, a link that does not even mention the name of the publication is not an adequate defense against plagiarism. The person also claims that the purpose of the website is not to receive clicks from Google’s search engine, but to test AI algorithms in multiple languages.

Reached for comment by email, Google declined to comment on Syrus. “We don’t comment on specific websites, but our updated anti-spam policies prohibit the creation of low-value, unoriginal content at scale for the purpose of ranking well on Google,” said Meghann Farnsworth, a Google spokeswoman. “We take global action against sites that violate our policies.” (Farnsworth is a former WIRED employee.)