[AI Rule Maker]⑤ Jungah Hwang: "Running a Marathon with Sandbags Makes No Sense... AI Regulations Should Be Delayed Three Years"

Complementary Measures Such as Safe Harbor Needed
Securing Sufficient Power and Quality Data Is Essential

"In the midst of intense competition between the United States and China, it is crucial for Korea to receive nationwide support in order to stand out and move ahead. However, putting up regulations first and then asking us to run is like running a marathon with sandbags strapped to your legs."

Jungah Hwang, a member of the Democratic Party of Korea, stated in an interview with The Asia Business Daily that amendments are necessary to relax certain regulations under the world’s first "Basic Act on Artificial Intelligence Advancement and Establishment of Trust" (AI Basic Act). Last year, Assemblywoman Hwang proposed a bill to postpone regulatory provisions for three years, which is currently being discussed within the Science, ICT, Broadcasting and Communications Committee. The AI Basic Act is currently in a state where the imposition of fines has been postponed for one year.

Assemblywoman Hwang remarked, "While drafting the AI Basic Act, we benchmarked the European Union (EU), but this June the European Parliament passed a motion by an overwhelming majority to postpone high-risk AI regulations until December 2027," and went on to point out, "Even advanced countries we are referencing have delayed all regulation until the end of next year, yet we are going ahead with implementation first." She emphasized, "This is not to say we should remove all regulations; for high-risk regulations, a pinpoint approach would suffice, but applying all regulations indiscriminately will make things very difficult for industry. Postponing regulation does not mean inaction. If companies have faithfully adhered to the standards set by the government, there should also be a 'safe harbor'—a complementary system to exempt them from liability for honest mistakes. The government should create an environment where companies can operate freely."

Jeonga Hwang, a member of the Democratic Party of Korea, is being interviewed by The Asia Business Daily at the National Assembly office. August 24, 2026. Photo by Hyunmin Kim

Speaking on the Yoon administration’s vision of making Korea one of the top three AI powerhouses, she said, "We are focusing on what we can actually do," highlighting AI data centers and physical AI as key strategies. Last year, she participated in the National Policy Planning Committee that shaped the blueprint of the current administration, handling national policy assignments involving AI, science and technology, aerospace, and defense industries.

Regarding AI data centers, Hwang commented, "With our experience in running systems and operating factories, such as through industrial engineering, I believe we have an advantage. While we may lack some element technologies, we have the ability to bring everything together and run things as a whole." On physical AI, she explained, "Physical AI refers to AI with a tangible presence—everything that moves, such as autonomous vehicles, drones, robots, and humanoids, will be equipped with an AI 'brain.' Since Korea is traditionally strong in automobiles and manufacturing, I think we can capitalize on that foundation."

However, she expressed significant concerns about the increase in related regulation. Assemblywoman Hwang said, "Physical AI is about moving from the software world into the physical world, where the challenges become much bigger. Even introducing a single robot leads to enormous regulation involving industrial safety, road traffic, and personal information, and there are currently no clear rules on determining responsibility in the event of an accident."

On this point, she added, "Before full implementation, my suggestion is to sufficiently conduct demonstrations and tests in a small, regulation-free environment called a regulatory sandbox before introducing AI into spaces where people are present. Since this is an issue involving multiple ministries, a consensus is needed for the regulatory sandbox, and I have prepared the legal basis necessary to enable this." The relevant bill was submitted to the National Assembly in June, titled "Special Act on the Promotion and Support of Physical Artificial Intelligence Development."

-What are the most frequent requests you hear from the industry and academia?

▲They are remarkably unanimous. There are two main requests: to innovate regulation to suit the AI era, and to provide comprehensive support for industry development. What frustrates companies the most is not a single large regulation, but rather a "web of regulations." Before the AIDC Special Act, even to build a single data center, permits and licenses were scattered across government ministries and local authorities, so it was impossible to know where a project might get blocked. Thus, the Special Act included an integrated permitting system centered on the Ministry of Science and ICT, a timeout system so that if a conclusion is not reached within a certain period, approval is granted automatically, exemptions for power grid impact assessments for data centers below a certain size outside the greater Seoul area, and the removal of commercial facility standards that required server facilities to have parking lots and art installations.

-You have described Korea's strongest area in AI as domain specialization. What does this mean in detail?

▲Take Kakao, for example. Kakao is not a company that directly develops every AI technology. Still, it delivers services to users and earns revenue that way. Korea is strong in creating specialized services. As a physics major, I believe continuous and stable investment in foundational technology is necessary. However, I also think that the actual value will come from domain-specialized areas such as medical AI or defense AI. Therefore, it is crucial to ensure that experts in each field are empowered to make effective use of AI.

-You have emphasized the importance of power and data. What is required to secure data?

▲Good models require good data, but in reality, high-quality data is often treated as an asset by companies and researchers who are reluctant to share it. That is why I proposed the National Research Data Act, which stipulates that research data generated by national projects should be standardized, managed by the government, and provided free of charge. There was significant opposition from civil society regarding the provision of sensitive data. However, we implemented multiple safeguards: sensitive data requires a separate process, and only data for which the proportion of state ownership exceeds a certain threshold can be provided to companies. This multi-layered approach was key to the bill’s eventual passage after much difficulty.

Hwang Jeonga, a member of the Democratic Party of Korea, is being interviewed by The Asia Business Daily in her National Assembly office. 2026.8.24 Photo by Kim Hyunmin

-Power remains a challenge. Do you still regret that LNG was excluded from the AIDC Act?

▲At the moment, it seems difficult to reintroduce it. The Ministry of Climate, Energy and Environment is currently saying that excess power can be used, but this is not accurate. Korea has serious power transmission network problems. Given the ongoing conflicts over transmission towers in neighborhoods, this is a very real issue. While the focus on renewables as the top priority is correct, it is almost impossible to attract large foreign tech factories without an LNG bridge.

-Why do you place such emphasis on power supply issues?

▲AI is like a brain, and ultimately, power is the key to supporting that brain. In my district of Daejeon, the electricity self-sufficiency rate is just 3%. Major cities like Seoul and Gyeonggi cannot even meet their current power demands, making it virtually impossible to attract AIDC. Thus, we must relocate centers to areas outside the Seoul metropolitan area, but to do that, we need a plan that includes how to provide electricity. In June 2024, when Seonghwan Kim, now Minister of Climate, Energy and Environment, was a three-term legislator, he once asked me, "What should we do about AI?" I told him then, "AI is an issue of power supply," and my view has not changed since.

-Are there any objections from labor unions regarding the introduction of physical AI?

▲Labor unions at Hyundai Motor have already gone on strike against the introduction of robots (Atlas). Just as with the Industrial Revolution, there is always labor opposition to changes that enhance efficiency, and I believe this is inevitable. However, this is a challenge that our society must address together.

-How should AI talent be developed?

▲Before AI, computer engineering was by far the top choice, but now only 2 out of 24 available positions are being filled in some programs. This is because companies are not hiring many new developers. As someone with a background in physics, I believe that strong fundamentals are important. However, Korea is more competitive in downstream services and domain-specialized AI than in element technologies. Therefore, while there must be ongoing investment in basic sciences, we also need more domain-specific experts.

-Do you see the decrease in new hires due to AI as a problem?

▲Even professionals like patent attorneys and accountants are not being hired as much these days because AI can do entry-level tasks better. Instead, companies are looking for more experienced mid-level professionals who can exercise human judgment. This leaves the junior generation, just beginning to build expertise, facing uncertainty about when and how they will grow. Such issues are common in periods of major transformation like the Industrial Revolution, but I believe our education system is itself in a transitional phase that requires change.

-You have worked to improve conditions for researchers.

▲One of the bills that recently passed the plenary session allows researchers at the four major Institutes of Science and Technology and government-funded research institutes to receive consulting fees or hold positions at companies established using technologies they developed. Previously, the system prevented this, resulting in loopholes like setting up 'figurehead CEOs.' Since commercialization of researcher technology is the fastest path forward, I believe this type of 'deep tech entrepreneurship' is essential. The science and technology R&D budget this year is 35.5 trillion won—one-twentieth of the total national budget—but among the 300 National Assembly members, very few examine it in depth. I believe the next two to three years represent a critical golden time for Korea.

-There is talk that the Science, ICT, Broadcasting and Communications Committee (SICT Committee) has become an undesirable committee to serve on. What are your thoughts?

▲That is true. This time, the People Power Party was given a quota of seven seats on the SICT Committee, but could only fill six. This reflects just how heavy the workload is. The science and technology R&D budget this year is 35.5 trillion won, which is one-twentieth of the total national budget of about 700 trillion won, yet few lawmakers look at it thoroughly. While everyone is keenly interested in broadcasting and telecommunications issues, there is a chronic lack of attention to science and technology legislation that will determine the country’s future.

-There are suggestions to separate the committee into science & technology and broadcasting. Do you agree?

▲Rationally, it makes sense. However, in practical terms, the main issue is that there are too few lawmakers with the expertise and dedication to delve deeply into science and technology legislation, despite the volume and importance of bills in this area. Rather than splitting the committee, I believe what is needed first is for the National Assembly as a whole to pay attention to science and technology in proportion to its budget scale.

This content was produced with the assistance of AI translation services.