| Mobile Web

AI safety differs from security; Korean-style technology can be competitive

AI safety should be approached through a separate evaluation framework, not as an extension of cyber security, Selectstar\'s AI safety team lead Kim Min-woo said. He cited risks such as hallucinations, hate, bias and harmfulness in generative AI. He said AI red-teaming tests models for unintended or harmful outputs, and that benchmarks must reflect local culture and goals. He also said AI agents using databases, RAG and plugins add complexity.