As South Korea's generative AI services market expands rapidly, business-to-consumer (B2C) payments are expected to exceed 1 trillion won this year. With ChatGPT drawing more than 5 million daily users and generative AI becoming an everyday digital service, user protection and operator responsibility are also coming into focus.

Data Mobile Index provided to DigitalToday on the 24th showed estimated payments for major generative AI services in South Korea, including Anthropic, OpenAI, Wrtn Technologies and Perplexity, totalled 962.3 billion won in January to August. With the total approaching 1 trillion won in eight months, an annual break above 1 trillion won appears certain.

The estimated payments are B2C transaction estimates calculated from a sample of domestic credit and debit card payments. Corporate account transactions and cash, gift certificates, simple payments and in-app payments were excluded, meaning the actual size of South Korea's generative AI market could be larger.

The market growth was led by Anthropic's Claude and OpenAI's ChatGPT. Anthropic's estimated payments for January to August were about 537.1 billion won, accounting for 55.8% of the total. OpenAI was about 393.8 billion won, or 40.9%. The two foreign operators accounted for 96.7% of the total estimated payments.

Domestic operator Wrtn Technologies accounted for about 23.7 billion won over the period, or 2.5%. Wrtn Technologies operates the generative AI portal Wrtn and the AI character chat service Crack, among others.

Claude's payment volume stood out relative to its user base. From June to August, Claude's daily average users (DAU) were about 280,000, about one-twentieth of ChatGPT's 5.34 million over the same period. But cumulative estimated payments in January to August were about 140.0 billion won higher for Claude than for ChatGPT. This was interpreted as reflecting service characteristics, including a high share of paid use for specialised purposes such as development, coding and work.

As the generative AI market grows, rules governing operator responsibility are expanding across the board. The Personal Information Protection Commission is overseeing personal data protection, the Ministry of Science and ICT is overseeing operator obligations under the AI Basic Act, and the Korea Communications and Media Commission is overseeing user protection for AI services, with each moving to improve systems.

In responding to personal data issues, insufficient cases have already been identified. On the 16th, the commission confirmed OpenAI had provided an email address for personal data inquiries to domestic users but did not reply, and demanded stronger management and supervision of its domestic representative. Wrtn Technologies was also included as a target for operational system improvements after it did not respond to inquiries on exercising personal data rights.

The AI Basic Act requires generative AI operators to notify users in advance that a service is AI-based and to mark outputs as results produced by AI. Foreign AI operators above a certain size must designate a domestic representative and report to the science minister.

The commission launched a second phase of the 'public-private consultative body for protecting AI service users' in May with 35 experts, and is discussing responses to illegal and harmful information within AI services and operator safety measures. Shin Young-gyu (신영규), director general of the commission's Broadcasting and Communications User Policy Bureau, said, "AI technology that is not ethically trustworthy can cause social distrust and hinder the growth of the overall AI industry." He added, "Technical safety measures to promote technology and protect users must be implemented together."

In the National Assembly, legislation to strengthen duties to protect children and adolescents is also being pursued. In June, Cho In-cheol (조인철), a lawmaker from the Democratic Party, introduced an amendment to the Act on Promotion of Information and Communications Network Utilization and Information Protection requiring providers of conversational and generative AI to verify age, obtain consent from legal representatives and implement protective measures for risk information such as suicide, self-harm, misuse of narcotics and sexual exploitation. Cho stressed, "As AI has entered deeply into children's daily lives, the stages of children's development and safety must be given top priority from the service design stage."