[메트로신문] The shockwaves caused by the Chinese AI startup DeepSeek are sweeping across the globe, with security issues such as personal data breaches emerging as a new concern.

Some countries have banned the use of DeepSeek due to the potential for personal data leakage, while others, particularly those in developed nations, are encouraging the active adoption of DeepSeek, driven by hopes of breaking the monopolistic AI technology regime established by advanced countries.

According to foreign reports, including Bloomberg, major global corporations and government agencies are blocking access to the AI inference model "DeepSeek R1" developed by DeepSeek.

DeepSeek R1 is a large-scale language model (LLM) trained on 671 billion parameters, and it has caused a major stir by delivering GPT-4-level performance at a low cost. It is particularly noted for its use of NVIDIA's H800 chip, which was released with reduced performance for the Chinese export market, but still manages to deliver performance close to that of GPT-4.

The U.S. government and Congress have taken a strong stance by fully banning the use of DeepSeek and are responding aggressively. The U.S. Congress has restricted the use of DeepSeek on public devices like phones and computers within Congress and has directed staff not to install it. The Congress has warned that "threat actors are already exploiting DeepSeek as a means to distribute malicious software and infect devices."

The U.S. Navy and Department of Defense have also blocked employee access to DeepSeek. It is reported that the Department of Defense took action after confirming that some employees had used DeepSeek, leading to the decision to restrict access.

At the corporate level, the movement to block DeepSeek is also spreading. According to network security firm Netskope, 52% of its clients have already fully blocked access to DeepSeek.

Nadir Izraeel, the Chief Technology Officer (CTO) of cybersecurity company Armis, stated, "About 70% of our clients have requested to block access to DeepSeek," adding, "Hundreds of companies, especially those linked to the government, are concerned about the potential data leakage to the Chinese government and vulnerabilities in personal data protection."

DeepSeek's privacy policy explicitly states that user data is collected and stored on servers in China, and any disputes related to this matter are subject to Chinese law. With this information now public, many companies and government agencies are blocking the use of DeepSeek's AI model R1 and its application programming interface (API) for security reasons.

Regulatory movements against DeepSeek are also gaining momentum in Europe. The Italian Data Protection Authority requested information regarding DeepSeek's handling of personal data, and as a result, DeepSeek's app was removed from Google and Apple app stores in Italy.

The Irish Data Protection Commission (DPC) also sent a letter to DeepSeek requesting information on how user data from Ireland is being processed. The letter expressed concerns about potential violations of the European Union (EU) General Data Protection Regulation (GDPR).

The French Data Protection Authority (CNIL) has also launched its own investigation into the functioning of DeepSeek's AI system and the potential data protection risks associated with it.

In contrast, India has embraced DeepSeek. Unlike the U.S. and Europe, which have adopted a strict regulatory stance, India is eager to actively utilize the technology. Ashwini Vaishnaw, India's Minister of Electronics and Information Technology, praised DeepSeek's technology and announced plans to host the DeepSeek model on servers within the country.

Moreover, India has been inspired by DeepSeek's low-cost AI and announced plans to develop six of its own foundational AI models by the end of the year. The government plans to expand AI usage in various fields such as agriculture and climate change through state support.

An Indian government official stated, "The foundational AI models developed in India will be able to compete with the top models in the world," adding, "There will be active government support to enhance AI technological competitiveness."

Experts point out that the technological innovation of DeepSeek R1 is sparking both expectations and concerns. An AI security expert commented, "The fact that DeepSeek R1 achieves GPT-4 level performance at a low cost is a technological leap, but it also requires a cautious approach in terms of data security and personal data protection." The expert further stated, "In South Korea, we must not only prepare countermeasures for the security threats posed by DeepSeek but also seek a balanced strategy that maximizes the potential of technological advancements."

ChatGPT를 사용하여 번역한 기사입니다.