Korea's new AI ethics principles spark industry concerns over data disclosure, regulatory risks
The government's voluntary AI framework promotes safety, transparency and accountability, but companies worry it could shape future regulations and expose competitive training data.
A visitor speaks to a humanoid robot built by the Chinese AI company AgiBot at AI Summit Seoul & Expo 2026, held at Coex in Gangnam District, southern Seoul, on Aug. 19.WOO SANG-JO
The government approved a set of ethical principles to address concerns about the safety and reliability of AI on Monday.
While the principles are intended as voluntary norms to help prevent problems that could arise as AI advances, some in the industry say the guidelines mustn’t function as regulations that hinder industry growth.
The framework, approved at a written session of the 12th Science and Technology-related Ministers' Meeting, sets common standards for AI development and use and assigns roles to providers, users, government and society. It addresses issues including overreliance on AI, changes to jobs, copyright infringement and plagiarism.
The principles are not legally binding and carry no penalties for noncompliance. The government describes them as voluntary norms rather than regulations.
The final version followed consultations with industry actors, academia, civil society and the public that began last December.
The framework is built around three overarching values — human dignity, the common good of society and sustainability — and seven principles intended to put those values into practice.
Human dignity calls for people to be treated as ends rather than means. The common good of society seeks to extend AI's benefits beyond individuals to the whole of society. Sustainability calls for those benefits to be shared with future generations while preserving the social and environmental foundations of human life.
The seven principles are human-centeredness, privacy protection, fairness and inclusiveness, accountability, safety, reliability and transparency. They call for human oversight, protection of personal data, safeguards against discrimination, clear responsibility for harm, risk-based safety measures, dependable performance and disclosure of how AI is used and where its limits lie.
In particular, under the transparency principle, providers are encouraged to disclose necessary information when copyrighted or other protected material is used in AI training or in products and services. The information should cover where the material came from and how it was used.
A staff member demonstrates a four-legged robot at the Smart Vision booth during the International Exhibition on Environmental Technology and Green Energy 2026 at Coex in Gangnam District, southern Seoul, on May 20.NEWS1
"We will support the ethics principles so they take root as a practical standard that raises social trust in AI, promotes responsible development and use and helps spread its benefits," Deputy Prime Minister and Minister of Science and ICT Bae Kyung-hoon said.
The government said it will prepare case-based guidance, sector-specific voluntary checklists and ethics education materials to help organizations apply the principles.
Some industry sources said the transparency language could create pressure on companies even though the framework is voluntary.
"Disclosure of AI training data is still being debated worldwide, so it should not be included in the new principles," a team leader at an AI company said. "It could also influence the direction of AI-related regulation now being discussed in Korea."
The government describes the framework as a voluntary norm with no legal force and says it does not replace legislation.
"Training data is part of an AI company's competitive edge," another industry source said. "Including disclosure in the principles could create difficulties for companies."
This article was originally written in Korean and translated by a bilingual reporter with the help of generative AI tools. It was then edited by a native English-speaking editor. All AI-assisted translations are reviewed and refined by our newsroom.
