[Korea SMEs and Startups Agency]
[Korea SMEs and Startups Agency]

Agency moves to manage AI risks including bias, privacy breaches and security threats

NIA grants 'AI Ethics Compliance' mark; staff training conducted

Ethics and safety standards to apply across all stages of service planning, development and operation

The Korea SMEs and Startups Agency has established an internal policy requiring staff to personally verify all work outputs generated by artificial intelligence. The move aims to clarify accountability for errors, bias and data leaks as public institutions expand their use of AI across operations.

The agency announced Tuesday that it had proclaimed its "Korea SMEs and Startups Agency AI Ethics Charter" and received the "AI Ethics Compliance" certification mark from the National Information Society Agency (NIA).

The NIA grants the AI Ethics Compliance mark to institutions that meet set requirements for the ethical development and use of AI and for strengthening organizational accountability. The mark confirms that an institution has put in place ethical principles, training and a practical compliance framework — it does not certify the performance or safety of any specific AI service.

The agency said it established the ethics management framework to prevent bias, privacy violations, security threats and reduced transparency that could arise as AI transformation expands across the organization.

The AI Ethics Charter, adopted this month, sets out six principles: human-centered AI use; fairness and non-discrimination; transparent operation and adequate explanation; personal data and data protection; safety and reliability; and strengthened accountability for AI use.

The agency has posted ethics charter posters in offices and consultation spaces across all departments and conducted "AI Ethics and Responsible AI Use" training for employees handling related work. The training covered precautions for using generative AI, measures to protect personal and internal information, procedures for verifying bias and errors in AI outputs, and compliance requirements on copyright and security.

Staff are required to verify the accuracy and appropriateness of AI-generated results before using them in their work, rather than applying the outputs directly. The intent is to use AI as a tool that assists human judgment — not replaces it — in supporting small and venture businesses through loans, guarantees and other programs.

"AI transformation can earn public trust only when ethical principles such as fairness, transparency and safety take root across all operations," said Lee Byeong-cheol, vice president of the Korea SMEs and Startups Agency. "We will spread a culture of responsible AI use to the small and venture business sector as well."


hong@heraldcorp.com