A new study by Rankfor.AI reveals that AI recommendations systematically hide brands based on just one word in a user query: husband, wife, or partner.
In a controlled experiment of 299 identical gift-recommendation prompts, researchers changed only the gender wording. Everything else stayed the same. The outcome was striking and statistically undeniable.
One example stood out immediately: Amazon’s Kindle appeared in 100 per cent of “wife” gift recommendations and 0 per cent of “husband” recommendations, despite being a widely gender-neutral product.
“This isn’t a bug or random noise,” said Dmitrij Zatuchin, CEO of Rankfor.AI. “It’s category gatekeeping. AI systems are deciding which brands belong to which gender — and excluding them completely from other searches.”
Key findings:
AI recommends 41–61 per cent fewer brands for wife-framed queries than for husband-framed queries across major models
33 well-known brands are gender-locked, appearing only in one type of query and never in others
So-called gender-neutral partner queries are not neutral: they share 69 per cent overlap with wife queries and only 38 per cent with husband queries
Statistical testing confirms the pattern is not random (χ² = 137.32, p < 0.001)
Crucially, the bias is not about repeating stereotypes. Instead, AI removes entire categories of products before recommendations are generated. While recommendations for women remain diverse, the total number of brands considered is cut roughly in half.
“AI isn’t saying a brand is ‘for women,’” the report concludes. “It’s saying that certain products don’t exist at all for that gender. That’s far more damaging.”
As consumers increasingly rely on AI for shopping and discovery, brands may be losing visibility to half the market without knowing it. A brand can dominate one type of query and be completely invisible in another, making standard AI visibility checks misleading.
The study also warns that brands positioning themselves as inclusive or gender-neutral may unintentionally disappear from male-framed searches altogether.
“If a flagship product like Kindle can be gender-locked by AI,” the researchers note, “no brand should assume neutrality without testing it.”
Share