Advertised Finding Jeonse Rentals, But Unable to Access Real Estate Platform
Due to Naver Restricting Crawling Access to Real Estate, Cafes, and Blogs
AI Blocked from 'Irreplaceable' Data, Serving as Catalyst for Fair Value Discussions
The 'data barrier' set up by Naver Corporation has blocked ChatGPT. Although OpenAI advertised ChatGPT finding a new home as its first campaign in South Korea, the service was found to be incapable of accessing property listings on Naver Real Estate, which anyone can easily check.
The issue of data access amid competition in artificial intelligence (AI) services has surfaced through real estate information. This is expected to spark broader societal discussion regarding the fair value of data moving forward.
According to search results tested by Electronic Times on September 2 based on the scenario in OpenAI's first Korean commercial released in July, ChatGPT did not cite data from Naver Real Estate, considered the largest platform of its kind in the nation.
In the advertisement, a couple asked ChatGPT to find a house in the Sinchon area with a park suitable for walking their dog. ChatGPT recommended neighborhoods that met the criteria, but no scene showed it retrieving actual listings. When a reporter asked the same question directly, the system returned answers utilizing listing information from platforms like Karrot Real Estate and Ziptoss, excluding Naver. Even when requested to “search again based on Naver Real Estate,” the system failed to cite Naver Real Estate data.
This is because Naver is restricting data access for major AI bots. Naver has implemented the Robots Exclusion Protocol (robots.txt) across its services, including real estate, cafes, and blogs, blocking crawling by major AI bots such as OpenAI and Anthropic. The robots.txt protocol is an international standard for specifying access permissions for web crawlers. Citing this, ChatGPT answered, “Currently, through web search alone, it is impossible to retrieve and recommend real-time listing lists from Naver Real Estate.”
Regarding this policy, Naver explained, “The primary purpose is to protect user-generated content (UGC) from unauthorized harvesting by third-party services.”
Such data barriers are expected to serve as tools for securing competitive advantages in AI services and keeping competitors in check.
Property listing data is a key asset allowing Naver to differentiate itself from global Big Tech in the domestic real estate AI competition. Unlike shopping and travel data, which involve active external integration, it is difficult to secure alternatives for real estate listings where recency and accuracy are critical. In fact, Naver integrated a “Real Estate Agent” feature powered by property listing data into its AI search service, AI Tab, late last month.
Naver's data barrier policy is expected to expand beyond a conflict over openness versus blocking into discussions surrounding the 'fair value of data.' This comes as the importance of high-quality data determining AI performance grows, while an ecosystem that properly values such data has yet to take root.
Previously, global AI firms broadly trained on and utilized publicly available web data. Conversely, among platform and content providers that invested capital and infrastructure to build data, vigilance against unauthorized data usage by AI is growing.
Indeed, cases of setting robots.txt rules to block crawling of accessible web content are on the rise. Experts point out that without standards for paying fair compensation, the data ecosystem will ultimately contract, triggering a vicious cycle in which AI companies fail to secure high-quality data.
Lee Seong-yeob, professor at Korea University, said, “As AI model performance becomes more sophisticated, the direction of AI competition is shifting toward data competition.” Lee added, “Since agreeing on fair compensation for data among companies is difficult, the government needs to lead relevant discussions, including establishing standards for data value assessment.”
