AI Spera outlines AI threat prediction research at KIISC workshop
CEO Kang Byung-tak described training models on about 4.3 billion IP address assets collected through the Criminal IP platform.
AI Spera presented research directions for AI-based cyber threat analysis and prediction built on global attack surface data at the 2026 KIISC AI × RMF Workshop, held at Samsung SDS West Campus, according to dailysecu.com.
In the workshop's final session, which focused on cybersecurity-specialized AI models, AI Spera CEO Kang Byung-tak gave a talk on learning methods for global attack surfaces. He described using information on roughly 4.3 billion IP address assets accumulated through the company's Criminal IP cyber threat intelligence platform, together with their historical changes, as training data for AI.
AI Spera also presented a case in which past data on IP infrastructure used in recent cyber attacks on the financial sector was analyzed.
The company said it is researching an AI model that learns how attack infrastructure forms and changes, with the aim of identifying potential threats in advance.
AI Spera said it is taking part in the Naver Cloud consortium's project to develop a cybersecurity-specialized AI foundation model.
Quick answers
Where was the 2026 KIISC AI × RMF Workshop held?
It was held at Samsung SDS West Campus.
What data does AI Spera use for its AI threat research?
The company cited information on roughly 4.3 billion IP address assets collected through its Criminal IP platform, along with their historical changes.
What project is AI Spera involved in?
AI Spera said it is participating in the Naver Cloud consortium's project to develop a cybersecurity-specialized AI foundation model.