News

According to a report from DigiTimes, Samsung Electronics recently experienced a data leak incident with their chatbot, ChatGPT, within 20 days of implementation. The leaked data included semiconductor equipment measurement data, product yields, and other sensitive information, which were added to ChatGPT's learning database.

According to various Korean media outlets such as Economist and SBS, there were three incidents of ChatGPT misuse and abuse within Samsung. In these cases, internal employees used ChatGPT to ask questions and verify answers, which inadvertently became part of ChatGPT's learning database. This resulted in potential leaks of confidential information to other ChatGPT users. Two of the cases were related to semiconductor equipment, while the other involved meeting content.

The first incident involved an anonymous employee, referred to as A, from Samsung's Device Solutions (DS) department who downloaded faulty semiconductor equipment measurement database software. A then copied the problematic code to ChatGPT to seek help and inadvertently made the faulty code part of ChatGPT's learning data.

In the second incident, an anonymous employee, referred to as B, from the DS department entered program code related to equipment yields into ChatGPT and asked ChatGPT to optimize the code. In the third incident, another anonymous employee, referred to as C, from the DS department requested ChatGPT to create meeting records.

To prevent similar accidents from occurring, Samsung is developing measures to protect its confidential information. If similar incidents occur in the future, Samsung may cut off ChatGPT services internally. Additionally, DigiTimes reports that Samsung is considering developing their own AI services through the innovation center within their DS department.

Samsung has reminded its employees through an official announcement to be cautious when using ChatGPT and that all data entered into ChatGPT will be transmitted to external servers. As a result, sensitive information may be at risk of leaking out if ChatGPT learns Samsung's internal data.

Initially, Samsung banned the use of ChatGPT out of concerns about data leaks. However, since semiconductor technology is the foundation of generative AI, Samsung allowed the use of ChatGPT in their DS department from March 11, 2023, while the Device eXperience (DX) department still prohibits its use.