Google has been intentionally delivering misleading search results to companies that track search rankings and collect data for artificial intelligence, according to a new analysis published by Nozzle.
The analysis indicated that this deliberate manipulation, termed “poisoning” of Search Engine Results Pages (SERPs), aims to disrupt data collection efforts without indicating that the results provided are incorrect.
Nozzle reported that Google began displaying non-standard SERPs around Sept. 20, with the phenomenon initially affecting 25 percent of results. The rate has since stabilized to between 5 and 11 percent daily.
Cyrus Shepard, a figure mentioned in the analysis, noted that these poisoned SERPs introduce significant artificial volatility. Repeated searches for the same query showed only 20.7 percent retention of websites in poisoned results, compared to 88.5 percent for normal search outcomes.
The altered SERPs exhibited distinct characteristics, including a higher prevalence of user-generated content from platforms such as Facebook, Instagram, LinkedIn, Quora and Reddit. Conversely, they displayed less content from MapQuest, Yelp and YouTube.
Furthermore, poisoned SERPs featured fewer AI Overviews, appearing in 34.9 percent of cases compared to 54.2 percent in normal results. They also showed fewer “people also search for” features, at 52.7 percent versus 87.2 percent.
Conversely, the analysis found that poisoned SERPs displayed more image packs, present in 44.2 percent of results compared to 23.0 percent normally, and more forum blocks, at 27.0 percent versus 20.5 percent.
Nozzle indicated that the disruption is widespread globally but varies significantly across different countries, industries, types of searches and request routes. Advertising and shopping results, however, remained stable and unaffected by the alleged manipulation.
Source: Search Engine Journal
