[Asia Economy Reporter Choi Eun-young] The artificial intelligence (AI) chatbot 'Iruda' is facing controversy not only over sexual harassment and homophobia but also over privacy infringement. After it was revealed that conversations from users of the company's other services on KakaoTalk were collected for training, there has been a growing online movement demanding the suspension of the service.
According to the information technology (IT) industry on the 10th, users of the AI chatbot startup Scatter Lab's service 'Science of Love' are protesting, saying "It seems like my KakaoTalk conversations have been leaked." Personal conversations exchanged via KakaoTalk, including real names, bank account numbers, and account holders, were reportedly exposed as they were.
According to Scatter Lab, their recently launched AI chatbot service 'Iruda' was created based on KakaoTalk conversations collected from their other service, 'Science of Love.' 'Iruda' is said to have learned approximately 10 billion conversation data points exchanged between real couples through deep learning to develop the ability to converse like a real person.
'Science of Love' is a service where users provide their KakaoTalk conversation data, which the service provider analyzes to offer dating advice. Scatter Lab collected 10 billion KakaoTalk conversations during this process and promoted that this helped improve the AI's performance.
In response, users of 'Science of Love' who provided their conversation data to the company expressed dissatisfaction, saying, "I didn't know it would be used for such a service. It seems like personal information has been leaked." Although the service's privacy policy states that "collected information will be used for new service development," there was no notification about data usage when using the service.
As the controversy spread, Scatter Lab explained to users that "personal information was protected through anonymization" and apologized, saying, "We failed to specifically notify users about the use of data for new services." The company stated, "All data used for training has been de-identified," and added, "If users do not want their data to be used for training, it can be deleted. Before using conversation-related analysis services, we plan to add a notification and confirmation process regarding data usage."
Despite the company's explanation, noise surrounding 'Iruda' continues unabated. As allegations of sexual harassment, homophobia, and privacy leaks surfaced, netizens began a hashtag campaign (#Stop_Iruda_Operation) demanding the immediate suspension of the Iruda service.
Meanwhile, 'Iruda,' which launched on December 23 last year, is rapidly gaining popularity among people in their teens and twenties. As of early this month, the number of users surpassed 320,000, with about 210,000 daily users and a cumulative total of 70 million conversations. Among users, 85% are teenagers and 12% are in their twenties.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
