Last March, a troubling post surfaced on a dark web hacking community, known for its role in cybercrime. The post offered to sell the account credentials of a high-level manager from a major South Korean corporation in exchange for Bitcoin. If compromised, such an account could unleash a torrent of sensitive corporate data. Fortunately, before any transaction could occur, a specialized AI developed by South Korean startup S2W intercepted the threat. CEO Suh Sang-duk affirmed, “Upon verification, we confirmed the account belonged to a critical employee handling internal information. Swift action prevented potential havoc on manufacturing lines and payment systems through ransomware attacks.”
S2W made headlines last year with the launch of DarkBERT, the world’s first AI language model tailored for the dark web. This year, they introduced CyBERTuned, an AI model adept at both preventing and remediating cybersecurity threats, showcased at the prestigious North American Chapter of the Association for Computational Linguistics (NAACL) conference. Suh explained, “DarkBERT’s training on over 6 million dark web pages enables it to identify potential data leaks and decipher suspicious contexts associated with cybercrime.”
Emerging from a security research project at the Korea Advanced Institute of Science and Technology (KAIST), S2W was founded in 2018 by Suh and his university colleague Shin Seung-won, a cybersecurity professor. Suh explained, “Shin recognized the potential beyond academic research in analyzing dark web information, prompting us to start a company with three doctoral students.” With a background as a developer at TmaxSoft, an MBA from the University of Michigan, consulting experience at Boston Consulting Group, and a venture established at the Lotte Institute of Economy and Business Strategy, Suh laid the groundwork for founding S2W.
S2W’s partnership with Interpol (International Criminal Police Organization) in 2021 to analyze dark web threats underscores their global influence in combating cybercrime. Recognized by the World Economic Forum (WEF) among the top 100 tech firms, Suh highlighted, “Key South Korean institutions and major corporations like Samsung, Hyundai, and Naver rely on us to safeguard their digital assets.”
The rapid rise of S2W in the security market is partly due to the lack of other security companies capable of thoroughly tracking the dark web. Suh mentioned, “There are between 500,000 to 1 million active sites accessible on the dark web, and we are the only ones in South Korea with the technology to detect newly created sites.” The dark web uses a distributed network system that employs the computers of numerous anonymous users with special internet browsers, making its structure complex and difficult to trace. Suh added, “We have accumulated technical expertise by tracking and analyzing the major nodes (network connection points) that make up the dark web over a long period. This allowed us to apply this knowledge to our proprietary AI algorithms, enabling us to automatically detect and collect vast amounts of dark web data.”
The COVID-19 pandemic has further exacerbated cybersecurity threats, popularizing cloud computing and remote work, which have weakened corporate security and led to a surge in cybercrime. Suh said, “On the dark web, we discover crime information involving South Korean companies once or twice every two months.” Last year, virtual assets involved in cybercrime through the dark web amounted to $24.2 billion.
