NewsGuard introduced an artificial intelligence chatbot this week that asserts gender identities can range from seventy‑two to an unlimited number and rejects the notion of a binary biological sex, according to a review by the Washington Free Beacon.
Endorsements and source selection
The tool is marketed as a personal fact‑checking assistant and has received endorsements from The Atlantic and Snopes. Critics argue that the chatbot omits moderate or conservative viewpoints because it only draws on sources that NewsGuard rates sixty or higher on its reliability scale. A study cited shows right‑leaning outlets average a trust score of twenty‑six point four, below the threshold, while left‑leaning sites average sixty point six.
Content and omissions
When queried about biological sex, the chatbot dismissed the traditional two‑sex model as overly simplistic, citing a Huffington Post essay by a gender‑studies professor and a MedicineNet article that lists seventy‑two gender identities and separates gender from physical anatomy. It did not reference a Wall Street Journal opinion piece by biologists that argues sex is functionally binary.
Anthropic’s Claude responded that the standard biological answer is two, defined by gamete type, while noting complexity in physical expression.
Testing reveals errors
Testing found the chatbot incorrectly identified Dan Morenoff, director of the American Civil Rights Project, as Jeffrey Morenoff, a sociologist, produced inaccurate article dates and confused administrators at UCLA.
Company background and response
NewsGuard was founded by Steven Brill and Gordon Crovitz. The company sells its data to advertisers and artificial intelligence firms to combat misinformation, although no major AI firm has licensed its domestic source ratings. Chief operating officer Matt Skibinski acknowledged the tool is not flawless but defended its factual accuracy, stating he does not see how the data would be used for model training.
Critics warn that the source‑filtering approach may embed NewsGuard’s biases, such as rating Chinese state media higher than some conservative outlets, into AI systems presented as neutral. The company now distances itself from earlier statements about using its data to train AI models.