South Korea to launch sovereign cybersecurity AI model by year-end

South Korea plans to deploy a specialised, sovereign artificial intelligence model dedicated to cybersecurity by the end of this year, Science Minister Bae Kyung-hoon announced on Thursday, July 16.

The initiative is designed to counter escalating digital security threats and comes amid heightened domestic concern after Washington abruptly enacted export controls on highly advanced American AI models. Among the restricted technologies is Anthropic’s "Mythos 5," a cutting-edge model heavily utilised for detecting software vulnerabilities.

"We are pushing to create an AI model specialising in cybersecurity within this year by training our existing sovereign AI model on security-related data," Minister Bae said during a comprehensive policy briefing held at Cheong Wa Dae, which was presided over by President Lee Jae Myung.

Bae acknowledged that South Korea’s current sovereign AI capabilities are insufficient to adequately defend against rapid cybersecurity threats that are increasingly augmented by generative AI. He reiterated earlier calls for the country to embark on the long-term development of a proprietary frontier model capable of rivalling platforms like Mythos.

"In my humble opinion, South Korea could build one, too," Bae remarked.

However, the minister noted that significant infrastructure hurdles remain. He argued that the current volume of graphics processing units (GPUs) the ministry is able to allocate to domestic firms developing advanced foundation models is highly inadequate, prompting him to formally request additional budgetary support from the government.

Beyond infrastructure, the high-level policy briefing addressed legal frameworks to bolster national defences. Officials discussed strategies to institutionalise "white hacking"—the practice of legally accessing computer networks with the owner's consent to identify system exploits.

According to a ministry official, the Ministry of Science and ICT is preparing legislation that would establish legal parameters allowing ethical hackers to perform vulnerability tests on private companies under specific, pre-determined conditions, even without their explicit prior consent.

Bae concluded by noting that South Korea is currently ranked third globally in AI competitiveness by an international assessment agency. The minister expressed optimism that the country could climb to second place in the upcoming evaluation round scheduled for release in August.

By Tamilla Hasanova