Gwangju City Blocks 'Deepseek'... Strengthening Security Against Information Leaks

Access Blocked on Work PCs
Notification Sent to Districts, Corporations, and Agencies

Gwangju City blocked the Chinese generative artificial intelligence (AI) 'Deepseek' starting from the afternoon of the 6th. Photo by Gwangju City

The city of Gwangju announced on the 7th that it had blocked access to the Chinese generative artificial intelligence (AI) 'Deepseek' starting from the afternoon of the 6th. This measure was taken as a precaution to prevent risks such as the leakage of key information and data.

The city added the IP addresses that can access 'Deepseek' to the list of harmful sites, making them inaccessible, in accordance with the regulations restricting internet use under Article 47 of the National Information Security Basic Guidelines.

This action follows a request from the Ministry of the Interior and Safety on the 4th, urging central government agencies and 17 metropolitan local governments to exercise caution when using generative AI such as Deepseek and ChatGPT (GPT).

The city notified the five districts, public corporations, affiliated organizations, and funded institutions of the blocking of Deepseek in Gwangju, and instructed them to decide whether to participate in the block after their own internal review. In particular, security precautions were provided, advising users to refrain from using the service if there are concerns about personal information leakage or cyber security.

The city strictly prohibits the use of materials or images containing personal information when using generative AI programs, and instructed staff not to ask questions or upload information about policies under internal review to such programs. Additionally, it emphasized caution to ensure that the institution's key confidential information is not provided as data to generative AI models.

Park Hyemi, Director of Data Informatization, said, "Gwangju City plans to guide employees to use generative AI programs in compliance with legal and ethical standards," adding, "We will do our utmost to operate the 24-hour Cyber Incident Response Center to protect key information, digital assets, and data."