Michael Sellitto (마이클 셀리토), Anthropic's head of global policy, will visit South Korea next week. He is expected to meet the National AI Strategy Committee and discuss ways to cooperate on AI security.

On May 6, industry sources said the two sides plan to hold policy talks aimed at strengthening AI security, focusing on Mithos, Anthropic's cybersecurity-focused AI model.

Mithos is a model specialised in detecting software vulnerabilities and analysing multi-stage cyberattack simulations. Its ability to identify complex code flaws is stronger than that of general chatbot-style AI. It caused a shock after it became known that it found vulnerabilities that human security experts had missed for 27 years.

Citing concerns about potential misuse, Anthropic did not release Mithos publicly and instead launched Project Glasswing, a public-private joint security consultative body. About 40 companies, including Amazon, Apple, Microsoft, Google and Cisco, are participating, and Mithos is provided on a limited basis only to them.

The South Korean government is also seeking to join the project. Ryu Jae-myung (류재명), second vice minister at the Ministry of Science and ICT, said at the World IT Show on April 22, "We are sounding out South Korea's official participation in security discussions including Anthropic's Glasswing and OpenAI," adding, "There is a need to share information directly in some way."

Anthropic is also reported to want a meeting with Bae Kyung-hoon (배경훈), deputy prime minister and minister of science and ICT.

The actual meeting lineup has yet to be decided. The Democratic Party decided to nominate Lim Moon-young (임문영), the standing vice chair of the National AI Strategy Committee, as its strategic candidate in the by-election for the National Assembly seat in Gwangju Gwangsan-eul to be held on June 3. Industry sources are leaving open the possibility that Lim could attend in another capacity or send a representative.