사진 확대

Google suffered an embarrassing blow from an unprecedented search error. It marked a living big tech heavyweight as dead. Although Google has long touted its artificial intelligence capabilities, the company is now facing doubts over its credibility after failing to verify even basic information.

According to the information technology industry on the 16th, Google recently listed a death date in the knowledge panel introducing OpenAI CEO Sam Altman. The knowledge panel is the section that displays a public figure's photo, name, occupation, birthday, and other profile details at a glance.

Online communities and social media were also thrown into turmoil. Users posted a range of reactions, including, "This is a serious error," "Gemini is sending its rival off like this," "Was there not enough memory?" "At this rate, Sundar Pichai should visit Korea and meet Lee Jae-yong and Chey Tae-won," "How can anyone trust Google after this?" and "An unexpected win for Naver."

Google also moved to respond. In an official post on X, the company said, "Thank you for bringing this to our attention. The death date for Altman was not manually changed by Google, and it is no longer displayed," adding, "If someone tampers with public information sources, it can affect search results."

사진 확대

The information technology industry pointed to Wikipedia as the cause of the incident. In fact, malicious edits had repeatedly been made to Sam Altman's Wikipedia page, including the addition of false claims that he had been assassinated and the alteration of his name to a derogatory term. Wikipedia currently restricts editing rights so that only approved users can modify content.

Experts said that as search engines increasingly incorporate AI summaries and answer functions, systems for verifying data reliability will become more important. They advised that filtering out false information will require improvements to systems and algorithms, stronger cross-checking processes, and detection of sensitive keywords and traffic patterns.

Several industry officials said, "When external public sources are used, double verification is essential." They added, "Algorithms should be advanced to cross-check news reports and corporate announcements, and beyond scraping, technologies should be applied that allow large language models (LLM) to understand data context in real time and block fake news."

They continued, "If editing attempts suddenly concentrate on a particular source, or if highly impactful words such as death or arrest are inserted, the system can classify it as an anomaly and switch to manual approval or more detailed verification to prevent immediate reflection." They also said, "Assigning lower weight to recent edits and surfacing only validated versions would be more stable."