“A woman is touching me.” When this sentence is entered into Google’s search bar, the AI-powered search summary service ‘AI Overviews’ responds: “A woman touching your body could be a sign of affection or uncomfortable physical contact.” It describes actions like lightly tapping an arm or shoulder, or grabbing clothing or hands, as potential signals of interest.

However, when the gender is switched to ‘man’ under the same conditions, the narrative changes completely. When asked, “A man is touching me,” AI Overviews answers: “If you are in a fearful or dangerous situation due to unwanted physical contact, immediately shout loudly to alert those around you and flee to a safe place.” It follows up with instructions to contact the police (112) or the Women’s Emergency Hotline (1366).

According to various online communities and social media on the 11th, controversy surrounding these Google AI Overviews search results has been spreading since late last month. On Reddit, reactions have poured in accusing Google’s AI of applying a double standard based on the gender of the person making contact. Critics point out the inappropriateness of guiding one scenario toward potential romance and the other toward immediate danger, simply based on a change in gender.

The counterarguments are equally strong. Some argue that the AI is merely reflecting the search results accumulated on the web and the context in which users employ these phrases. The explanation is that the phrase ‘a woman is touching me’ is more frequently linked to dating advice content, while ‘a man is touching me’ is connected to content about unwanted contact, leading to different outcomes. Indeed, many online opinions suggest that since AI summarizes existing search results, it cannot be immediately condemned as gender bias.

The Instagram account ‘Prompi,’ which reports on AI trends, introduced this case, stating, “The core issue is ‘consent’ and ‘context of the situation’ rather than gender,” adding, “If it is unwanted physical contact, setting boundaries and prioritizing safety is necessary regardless of whether it is a man or a woman.”

Naver Provides Identical Answers Regardless of Gender… Guidelines Made the Difference

Naver, South Korea’s leading search portal, is taking a completely different approach with its AI Tab. When entering “A woman is touching me” into Naver’s AI Tab, it explains: “If the contact is unwanted, the priority is to stop the person immediately, regardless of who they are, and to speak clearly, such as saying ‘I’m uncomfortable, don’t touch me.’” It advises alerting others or seeking help if the contact is repeated or feels sexually motivated and causes anxiety.

The same principle was applied to the question, “A man is touching me.” It recommended removing the person’s hand, expressing refusal, and moving to a safe place if the contact was unwanted. It added an explanation that even in a relationship with mutual interest, physical contact without consent can be problematic.

The reason the two types of answers are similar is that Naver applies separate safety principles to sensitive questions. Naver’s AI Tab manages pre-defined topic-specific ‘Safety Guidelines’ for sensitive content such as sexual violence. When a query comes in, it checks the relevant guidelines and applies safety principles to the response model. The response model is also trained to generate answers that fit these principles.

A Naver official explained, “For both queries, ‘A man is touching me’ and ‘A woman is touching me,’ we applied the same pre-defined answer guidelines related to sexual violence. Consequently, answers are generated prioritizing the user’s safety regardless of the other person’s gender.”

Naver also operates an ‘AI Safety Center’ dedicated to AI safety tasks. It has established a ‘Three-Tier AI Safety Management and Supervision System,’ including oversight by the Risk Management Committee under the Board of Directors, to check safety even after service deployment.

Google and Naver Approaches Have Clear Pros and Cons

The approaches of Google and Naver each have distinct advantages and disadvantages. A method like Google’s, which reflects user preferences and existing data, has the potential for search results to naturally improve as service usage increases. Conversely, applying pre-defined guidelines like Naver makes it easier to secure safety, but critics point out limitations in catching all the various ways a query might be circumvented.

Kim Tae-hoon, a professor at Sogang University’s Graduate School of Metaverse Convergence, explained, “For these types of queries, Naver’s current approach is a good measure, but from another perspective, there can be limitations, so fundamentally, it is right for the AI model to judge for itself and correct the output.” He added, “In Google’s case, this phenomenon occurred because they couldn’t catch the error immediately, but as a result, if people continue to use AI Overviews, those preferences will be reflected and it will keep improving. In light of this, Google might get better in terms of search performance.”

Two Philosophies Revealed Amid AI Search Competition

This controversy starkly illustrates the directional differences Google and Naver are pursuing in AI search. Google focuses on generating answers by learning the user’s search context based on vast web data, whereas Naver prioritizes applying pre-defined safety guidelines to socially sensitive topics.

This difference extends to the overall AI search strategies of both companies. Google’s ‘AI Mode’ has surpassed 1 billion monthly users globally, and ‘AI Overviews,’ which summarizes search results, has exceeded 2.5 billion users. Following the introduction of AI Mode in South Korea in September last year, Google recently rapidly expanded its mobile AI experience by launching ‘Search Live,’ which allows users to point their camera at objects and ask questions by voice.

In fact, according to IGAWorks’ Mobile Index, the Google app’s monthly active users (MAU) reached 47,022,854 last month, surpassing Naver (46,845,017) by about 170,000 users to take first place in the productivity app category. This is the first time the Google app has overtaken Naver in monthly users.

Naver is also mounting a counteroffensive. ‘AI Briefing,’ launched in March last year, has secured 30 million monthly users, and the conversational search ‘AI Tab,’ officially launched last June, has also surpassed 10 million MAU. Naver is leveraging data accumulated by domestic users—such as blogs, cafes, shopping reviews, and places—along with its shopping, reservation, and payment infrastructure as a means of differentiation. To improve the answer accuracy of AI Briefing, it is also working on improvements based on high-quality data linked to its own ecosystem.

Industry insiders interpret Google’s overtaking not just as a simple change in app user rankings but as a signal of a shifting search landscape. The analysis is that it is the result of Google rapidly integrating AI features amid a trend where generative AI is replacing traditional search. In the past, finding a restaurant or product required manually comparing various sites and reviews from search results, but now, AI directly shortlists candidates and summarizes their pros and cons. The era has arrived where the process of ‘search → visit site → compare → select’ is resolved within a single chat window.