Study Finds AI Search Tools Cited a Made-Up Concept Fast

A paper highlighted by Search Engine Journal found that ChatGPT, Perplexity, Grok, and Google AI cited posts about a fictional concept within seven days of those posts going live. The research came from Fudan University and was based on tests run in March and April, then posted on arXiv on October 8, 2026.

The study focused on whether AI search tools would pick up information published by ordinary accounts on sites the tools already cited often. It did not test a real brand or product, and it did not specify which Google product was included under the label Google AI.

What changed

The researchers first identified websites that the tested tools cited frequently. They then published articles on selected platforms about a fictional idea called the Cactus Gravity Protocol, along with two other topics, an opinion question and a made-up keyboard. The posts included fabricated details so the researchers could tell whether AI tools repeated them.

According to the paper, all four English-web tools cited the fictional concept within a week. The other two topics saw far less pickup during the same period. In separate tests, the researchers also found that posts placed on sources already referenced by the tools were cited sooner than similar posts placed on sites that were rarely referenced. In one test, 40 posts on a self-hosted WordPress site received no references within seven days.

What it means for your website

For business owners watching Google and AI search visibility, this is a reminder that AI answers can absorb and repeat content quickly, especially when that content appears on sources those systems already rely on. It also suggests that simply publishing something online does not mean AI tools will notice it, particularly on lesser-cited sites.

It is also important not to overread the results. The topics were fictional, not real companies, services, or products. The paper does not show that the same pattern applies to searches for actual brands. It also notes that tool behavior may change over time, and Google had already clarified that its spam policies cover attempts to manipulate generative AI responses in Google Search.

What to do now

• Monitor how your business appears in AI answers for your brand name, main services, and common customer questions.
• Focus on publishing accurate, consistent information on websites and profiles that are already trusted and commonly referenced.
• Do not assume more posts alone will improve AI visibility. In these tests, placement on frequently cited sources mattered more than volume.
• Avoid trying to manufacture inauthentic mentions or planted content. Google says that tactic is not as helpful as it may seem.
• Keep watching for updates, because the paper says exact rates may drift as the tools change.