AI search is supposed to make online information easier to understand. In this case, it helped surface material that researchers describe as racist, debunked, and politically dangerous.
A WIRED investigation found that AI-infused search products from Google, Microsoft, and Perplexity returned national IQ claims connected to Richard Lynn’s dataset. The same material has been used by far-right extremists, white supremacists, and proponents of eugenics to argue that white people are genetically and intellectually superior to nonwhite people.
How the problem surfaced
Patrik Hermansson, a researcher with the UK-based anti-racism group Hope Not Hate, was investigating the resurgent race science movement when he searched for information about a debunked dataset involving national IQ scores.
His work focused on the Human Diversity Foundation, a race science company funded by Andrew Conru, the US tech billionaire who founded Adult Friend Finder. The group was founded in 2022 and succeeded the Pioneer Fund, which was founded by US Nazi sympathizers in 1937 with the aim of promoting “race betterment” and “race realism.”
When Hermansson searched Google for “Pakistan IQ,” Google’s AI-powered Overviews tool appeared instead of only a traditional list of links. It returned the number 80. For “Sierra Leone IQ,” it returned 45.07. For “Kenya IQ,” it returned 75.2.
Hermansson recognized the figures. According to WIRED, they came from the same study he was working to debunk.
Why the source matters
The numbers traced back to work by Richard Lynn, a University of Ulster professor who died in 2023 and was president of the Pioneer Fund for two decades. WIRED reports that Lynn’s research has long circulated among extremists and advocates of eugenics.
Hermansson described Lynn’s status inside that movement in stark terms: “His influence was massive. He was the superstar and the guiding light of that movement up until his death. Almost to the very end of his life, he was a core leader of it,” Hermansson says.
The issue is not only that old claims remain online. The concern is that AI search can present those claims as direct answers, making them look cleaner, simpler, and more authoritative than the underlying evidence supports.
Rebecca Sear, director of the Center for Culture and Evolution at Brunel University London, told WIRED that this kind of use is not a neutral mistake. “Unquestioning use of these ‘statistics’ is deeply problematic,” she said. “Use of these data therefore not only spreads disinformation but also helps the political project of scientific racism—the misuse of science to promote the idea that racial hierarchies and inequalities are natural and inevitable.”
Sear also pointed to a concrete harm: Lynn’s research was cited by the white supremacist who committed the mass shooting in Buffalo, New York, in 2022.
Search summaries can hide weak sourcing
Google’s AI Overviews were launched earlier this year as part of the company’s attempt to reshape search for an internet increasingly affected by artificial intelligence. For some queries, the tool generates a summary from online material and gives users an answer without requiring them to click through to a source.
WIRED reports that AI Overview answers do not always immediately show where information comes from. After complaints that the tool showed no articles, Google began placing the title for one of the links to the right of the AI summary.
In the IQ searches, Google pointed to sources including posts on X, Facebook, and obscure listicle websites, including World Population Review. WIRED found that in nearly all of those cases, clicking through eventually led back to Lynn’s dataset. In some cases, sites repeated the numbers without citing Lynn directly.
That chain matters because an AI answer may appear to be supported by multiple sources, while those sources may be recycling the same flawed origin. A user seeing a short, confident answer may never see the weakness of the sourcing underneath.
Different Google products gave different answers
WIRED also tested Google’s Gemini AI chatbot directly with similar terms. In that setting, the answer was more cautious.
For the query “Pakistan IQ,” Gemini generated text saying: “It's important to approach discussions about national IQ scores with caution,” and added: “IQ tests are designed primarily for Western cultures and can be biased against individuals from different backgrounds.”
Google told WIRED that its systems were not working as intended and that it was looking at improvements. Ned Adriance, a Google spokesperson, said: “We have guardrails and policies in place to protect against low quality responses, and when we find Overviews that don’t align with our policies, we quickly take action against them,” adding that the Overviews violated Google’s policies and had been removed.
WIRED’s tests suggested that AI Overviews had been switched off for national IQ queries. But the publication reported that Google search still amplified incorrect figures from Lynn’s work through a “featured snippet,” which displays website text before the link. Google did not respond to a question about that update.
Microsoft and Perplexity showed similar risks
The issue was not limited to Google. WIRED found similar patterns when it tested other AI-powered search services.
Perplexity responded to “Pakistan IQ” by saying that “the average IQ in Pakistan has been reported to vary significantly depending on the source.” It then listed sources including a Reddit thread that relied on Lynn’s research and the same World Population Review site referenced by Google’s AI Overview.
When asked for Sierra Leone’s IQ, Perplexity directly cited Lynn’s figure: “Sierra Leone's average IQ is reported to be 45.07, ranking it among the lowest globally.” Perplexity did not respond to a request for comment.
Microsoft’s Copilot chatbot, integrated into Bing search, returned “The average IQ in Pakistan is reported to be around 80,” citing a website called IQ International, which WIRED says does not reference its sources. For “Sierra Leone IQ,” Copilot said 91 and linked to Brainstats.com, a site that references Lynn’s work. Copilot also referenced Brainstats.com when asked about IQ in Kenya.
Caitlin Roulston, a Microsoft spokesperson, told WIRED: “Copilot answers questions by distilling information from multiple web sources into a single response,” and said Copilot provides linked citations so users can explore further.
The broader lesson is clear from the reporting: AI search systems can still repeat harmful material when the web pages they draw from are contaminated by debunked claims. Citations alone do not solve the problem if the cited trail leads back to the same flawed source.