
Upload photos to an internal employee-generated AI service... There are no additional leaks
AI-generated summary
During the process of developing and testing the AI system, the Public Procurement Service suffered a leak when an internal employee uploaded a photo to verify the identity of an evaluator to an externally generated AI service.
The Public Procurement Service announced on the 10th that identity verification photos of 189 evaluators who participated in online evaluations were leaked during the development of an artificial intelligence (AI) system.
This leak was not caused by external hacking, but occurred during the process of developing and testing an AI system to improve online evaluation work, when an internal employee uploaded a photo to verify the evaluator's identity to an externally generated AI service between July and October.
The Public Procurement Service explained that the target of the leak was 189 evaluators who participated in the online evaluation, and that the leaked items were limited to photos used to verify identity and prove the fairness of the evaluation.
The Public Procurement Service further added that it was found that the photos did not contain other personal information, such as names, and that unique identification information and financial information, such as resident registration numbers and account numbers, were not leaked.
On the 6th, when the Public Procurement Service finally confirmed the leak of photos of the evaluators, it immediately reported it to the Personal Information Protection Committee. Afterwards, it disclosed the leak on the Public Procurement Service website in accordance with the procedures stipulated by relevant laws and regulations, and is providing individual notification to all relevant evaluators.
A dedicated response team (TF) has been formed to investigate the leak and its spread, and an official request has been made to the generative AI service provider to delete the photo.
The Public Procurement Service plans to continue to check for additional damage and take necessary protective measures.
In addition, we plan to make every effort to prevent recurrence by improving the overall security system for AI use, such as blocking unauthorized transmission of personal information with externally generated AI and strengthening security training for all employees.
Baek Seung-bo, head of the Public Procurement Service, said, "We deeply apologize for failing to thoroughly protect the valuable photo information of the evaluation committee members," and added, "We will fully fulfill our responsibilities for damage relief and follow-up measures, and do our best to prevent similar incidents from recurring by fundamentally improving the information protection and security system."
AI outlook — possibilities, not facts
The Public Procurement Service will improve the security system utilizing AI and implement measures to prevent recurrence.
Likely · Within weeks

Samsung SDI won a CES 2027 Innovation Award for its SDI U8A3-UPS battery system, designed to provide backup power for AI data centers. The prismatic lithium manganese oxide-based system offers 33 percent greater space efficiency than previous products and passed an indoor large-scale fire test by UL Solutions, becoming the first product worldwide to meet all evaluation criteria.

Kolmar Korea won three CES 2027 innovation awards for two AI-powered beauty devices, including Best of Innovation in beauty tech for ChromaFlow, which creates personalized lip colors in about one minute. The company also won AI category awards for ChromaFlow and LumiJet Pro, marking its second consecutive year winning the top Beauty Tech award.

South Korean healthcare appliance maker Ceragem won nine CES 2027 Innovation Awards for eight products, including AI-powered home healthcare systems, smart cribs, and wellness furniture, marking its fourth consecutive year of wins and bringing its total to 30 since 2024.

The Ministry of Land, Infrastructure and Transport will improve the railway vehicle type approval system to strengthen inspection of high-risk parts and reduce duplicate inspections, and support exports by gradually revising domestic technical standards in line with European standards (TSI). Technical standards will be revised in stages starting from 2028.

A research team from POSTECH and KAIST has developed a technology that uses the screen reflection of foldable phones to prevent people around you from peeping. It can be implemented simply by introducing software without any additional devices, and the numbers cannot be read even if they deviate from the user's line of sight by just 7.5 degrees. Received an excellence award in the demo category at ACM UIST 2026.

DX-M2, the next-generation AI semiconductor produced by DeepX using Samsung Electronics' 2-nano process, received the CES 2027 Innovation Award. The goal is to implement generative AI inference in the device itself with about 5 watts of power, and the plan is to begin initial mass production in the fourth quarter of next year and begin full-scale mass production in 2028.