The Telecommunications Technology Association (TTA) said on Monday it has set up a dedicated project group (PG507) to standardise “AI security” to ensure the reliability of artificial intelligence models and protect data.

The project group will pursue domestic and international standards focusing on information security technologies for AI systems, AI privacy protection technologies, AI security evaluation technologies, generative AI security technologies and agentic AI security technologies.

AI has recently moved beyond simple automation and is rapidly advancing, led by generative AI. It is converging with various industries, including real-time environmental perception and decision-making for self-driving cars, personalised educational content generation and intelligent robot control, helping drive the creation of new industries.

As access to AI technology increases, security threats such as cybercrime and privacy violations that misuse it are also rising. TTA said the project group plans to develop AI security standards that consider new types of security threats such as model theft, prompt injection and adversarial attacks, as well as AI-related laws and regulations in each country.

TTA Chairman Seung-hyun Son (손승현) said the project group would contribute to building a standards ecosystem for fostering the AI industry and ensuring technology reliability. He said he expects it to provide a basis for reflecting South Korea’s advanced technologies in international standards by cooperating with domestic and international AI standardisation bodies.