[Asia Economy Reporter Koo Chae-eun] Scatter Lab, the developer of the AI chatbot 'Iruda,' has decided to discard Iruda's database (DB) and the deep learning conversation model used for training.
Scatter Lab plans to proceed with the disposal of Iruda's DB and deep learning conversation model immediately after the investigations by the Korea Internet & Security Agency (KISA) and the Personal Information Protection Commission are completed.
On the 15th, a Scatter Lab official stated, "Considering users' concerns, we have decided to discard the entire database and deep learning conversation model of the AI 'Iruda.'"
The existing data collected with user consent from "Science of Love" and "Text@," will be deleted upon requests from users who do not wish their data to be used. This data will no longer be utilized in future deep learning conversation models.
Scatter Lab explained that Iruda's DB has undergone de-identification procedures and consists of individual and independent sentence units. There is no personally identifiable data. Furthermore, the deep learning conversation model learns only conversation patterns based on de-identified data, and since the AI stores data as vector values, there is absolutely no risk of personal information leakage.
Related follow-up measures will be announced through each application’s notifications. In the future, the procedures for obtaining consent for personal information collection and use will be strengthened during new registrations and service usage.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
