[Mega Project Success Factors]③ Ecosystem Over Technology...Urgent Need for Regulatory Reform in Physical AI

Existing Laws on Safety Cannot Be Easily Applied
Barriers Such as Data Collection and Personal Information Must Be Removed for Growth

Hyundai Motor Group Has Secured World-Class Technology
Regulatory Innovation and Expanded Testbeds Must

The final and most critical piece in the government's ongoing mega-project initiative is the establishment of a physical artificial intelligence (AI) ecosystem. While domestic companies such as Hyundai Motor Group possess advanced robotics technologies and manufacturing capabilities, securing global leadership will not be possible through technical prowess alone.

Experts have assessed that outdated regulations must be dismantled, large-scale demonstration environments and robust data utilization infrastructure must be established, and a comprehensive talent and investment ecosystem must be built within the next three years. The ultimate success or failure of the mega-projects will hinge on how quickly the government can provide a foundation to connect corporate technologies with real-world industries and markets.

'From Moving Machines to Judging AI'...Urgent Transition of Laws and Systems Needed

Regulatory reform and the overhaul of legal frameworks were cited by experts as the most pressing issues. Currently, the Occupational Safety and Health Act and the Product Liability Act are designed under the premise of human-operated machines, and are not readily applicable to autonomous vehicles and humanoid robots equipped with AI capable of independent judgment. Moreover, the boundaries of liability in the event of an accident—whether it rests with the manufacturer, AI developer, operator, or end user—remain unclear.

Yoo Seunghoon, professor at the Department of Future Energy Convergence at Seoul National University of Science and Technology, stated, "The current legal system assumes 'machines that operate in a predetermined manner,' making it ill-suited to include physical AI. We must develop a comprehensive negative regulatory safety certification framework—allowing most activities while specifically prohibiting exceptions—and institute a new liability insurance system to cover accidents resulting from AI decision-making."

On the other hand, Lee Byungheon, professor at the School of Business Administration at Kwangwoon University, maintained a cautious stance regarding preemptively legislating safety standards and liability structures. "Determining the safety certification and liability structure for physical AI entirely through existing laws from the outset is both unrealistic and ineffective," he said. "It would be more efficient to initially allow limited commercialization through a regulatory sandbox and, based on accumulated case studies, gradually develop the necessary legal systems."

Yang Junseok, professor of economics at Catholic University, added, "Without proper insurance and liability frameworks, the autonomous vehicle industry cannot grow. Instead of stifling experimentation through excessive pre-regulation, it is preferable to expand demonstrations on the principle that companies assume civil liability."

Data collection also poses challenges. Industry stakeholders stress that the strength of physical AI technology hinges on data collection. In the process of gathering data to train robots, sensitive personal information—such as faces, license plate numbers, and movement paths—may be inadvertently captured.

Domestically, according to the Personal Information Protection Act, any video footage intended for AI training must be processed to prevent personal identification, resulting in dramatically escalating costs and manpower requirements for pseudonymization, validation, and data management as data accumulates. To address this, experts unanimously pointed out the need to lower the regulatory barriers for personal information, specifically through mechanisms like regulatory sandboxes.

Boston Dynamics Humanoid Robot Atlas. Hyundai Motor Group

Hyundai Already Prepared...Ecosystem Readiness Will Further Accelerate Progress

Experts agree that Hyundai Motor Group, which is a crucial pillar of the mega-project, has established the nation's most advanced infrastructure for implementing physical AI.

Professor Yoo stated, "Hyundai has acquired world-class actuator and robotics technology via Boston Dynamics, and is accumulating vast stores of behavioral data in smart factories such as the Singapore Global Innovation Center (HMGICS) and in its dedicated electric vehicle (EV) plant in Ulsan. For AI to function effectively in the physical world, high-quality behavioral data is essential, and Hyundai is the company most efficiently equipped to collect and utilize such data."

Professor Lee added, "Korean companies have the combined advantage of manufacturing data and expertise in robotics hardware design and production. If the mega-project establishes further AI infrastructure such as data centers, their competitiveness will only increase. However, the capability to swiftly develop 'Large Action Models'—which enable an understanding of the physical world—remains an urgent priority."

Professor Yang remarked, "Korea already has one of the world's highest rates of industrial robot adoption and is highly competitive in mechanical engineering. Hyundai, too, is advancing a strategy that aims not only to deploy robots for its own production, but also to nurture humanoid and autonomous technologies as future export industries."

In summary, experts share the consensus that while companies are largely technologically prepared, rapid growth of the domestic physical AI ecosystem is contingent on dismantling regulatory and institutional barriers.

"There Are Limits Inside the Factory"...Large-Scale Demonstration Environments Needed

To accelerate progress for physical AI firms—including those participating in the mega-project such as Hyundai Motor Group—regulatory innovation and expanded testbeds were cited as essential.

Physical AI can only build competitiveness when it learns from diverse real-world variables not just within factories, but also at ports, logistics centers, and urban environments. However, in Korea, various regulations continue to restrict open testing environments.

Professor Yoo emphasized, "For the government's data factory initiative to have practical impact, it must go beyond simply providing space. There is a need to create open innovation zones where laws like the Radio Waves Act, Personal Information Protection Act, and Road Traffic Act are temporarily suspended—only then can significant commercialization data be obtained."

Professor Lee noted, "Currently, Korea severely lacks appropriate demonstration environments. In conjunction with the '5-Pole-3-Special-Zone' regional growth strategy, local governments and the private sector should jointly establish large-scale testbeds across different regions."

Likewise, Professor Yang highlighted, "Just as new drugs must pass clinical trials, the completion of physical AI also demands real-world validation. Companies should be allowed to expand practical testing, provided they bear responsibility for any accidents."

Additionally, Professor Yoo stressed, "A reliable national power grid and energy supply system must be established to support the physical AI ecosystem. Stable power should be delivered to data centers and testing campuses, while energy strategies—such as expanding carbon-free power from offshore wind and hydrogen—should be put in place."

Talent Matters More Than Tax Incentives...Clusters Without Skilled People Have Limits

Experts have also identified talent development and industry ecosystem formation as another core challenge for the physical AI sector.

Professor Lee stated, "We need to nurture startups and ventures that develop advanced technologies, foster talent to lead them, attract globally skilled individuals, and build collaborative ecosystems between large corporations and SMEs."

A humanoid robotics expert at the Korea Institute of Science and Technology (KIST) commented, "Currently, most AI and robotics research talent is concentrated in the Seoul metropolitan area, especially Seoul and Pangyo. While data centers may benefit from being located in regional areas due to land and power advantages, relocating R&D organizations is difficult due to challenges in talent acquisition."

He continued, "It’s common for companies to place factories and production facilities in provincial regions while maintaining research institutes in Seoul. To successfully establish new AI and robotics clusters, not only land and power but also living and educational infrastructure and talent recruitment strategies must be developed."

Opinions differed on methods of support. Professor Yoo advocated for designating core technologies and capital expenditure (CAPEX) for physical AI as national strategic technologies to expand tax credits. In contrast, Professor Lee viewed R&D investment funds and venture capital as more effective than tax incentives for Startup-centered technology development. Professor Yang also said that easing regulatory restrictions on data collection and experimentation should be a higher priority than tax relief.

Ultimately, experts concurred that the success or failure of mega-projects will not be decided solely by the technological capabilities of any one company. With Hyundai Motor Group laying the groundwork for physical AI through Boston Dynamics, smart factories, and manufacturing data, the competitiveness of Korea's physical AI industry depends on how swiftly the government can establish regulatory innovations, safety systems, large-scale demonstration environments, an investment ecosystem, talent acquisition, and power infrastructure.

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