SANDS Lab sets a benchmark for security innovation with generative AI and datasets

SANDS Lab sets a benchmark for security innovation with generative AI and datasets

• AI advances usher in an era of growing importance for AI security
• Combining CTI with generative AI, LLMs and RPA
• Strengthening global technology development with generative AI specialists

Cyber threat intelligence specialist SANDS Lab is steadily evolving as an AI security company aligned with changing times through its strategy for generative AI and AI training datasets.

CTX for GPT, its generative AI-based intelligence service, recently attracted attention after joining the GPT Store. Cyber threat intelligence is the foundation that enables these AI security innovations.

Intelligence services analyze existing and potential threats and trends, providing customers with findings drawn from vast historical and current data to help them prepare. Through this work, SANDS Lab has collected and analyzed proprietary big data now measured in petabytes.

Figure 1. CTX, a next-generation AI-based cyber threat intelligence service

A collection or grouping of specific data is called a dataset. Training AI requires high-quality structured datasets. SANDS Lab’s data include diverse threat group and campaign information, helping AI algorithms handle more realistic and varied situations. It is no exaggeration to say that datasets determine the weaknesses and limitations of AI.

SANDS Lab has undertaken KISA dataset projects since 2021. Supplying datasets to government and public organizations over multiple years demonstrates the accuracy and consistency needed to support intelligent cyber defense and AI security product development, as well as future generative AI research and other applications.

Figure 2. Overview of SANDS Lab’s AI datasets

Building on its experience and datasets, SANDS Lab has successfully integrated generative AI and GPT to pursue innovation that sets new standards for security. Generative AI analyzes large volumes of internal network traffic, files and campaigns in real time and detects threats, while GPT improves natural-language understanding for more proactive analysis and response.

Figure 3. CTX for GPT combines generative AI with cyber threat intelligence

More recently, SANDS Lab developed SANDY, a lightweight large language model (sLLM) deployed on premises so companies can operate and control it directly within their own infrastructure.

Designed for privacy and compliance, SANDY provides large-scale language processing while running inside the corporate network, significantly reducing leakage risks. It acts as an AI cybersecurity expert, taking over tasks such as report summarization and editing and internal searches for hacking response information.

Figure 4. SANDY, an on-premises generative AI cybersecurity model

CEO Kim Ki-hong said, “Public and private efforts to use CTI and AI training datasets effectively can drive technological progress and innovative industry outcomes. To help everyone anticipate and counter the latest threats and stay protected from AI-enhanced attacks, generative AI, LLMs and RPA are essential to better detection and response. We will accelerate globally competitive development through cooperation with Korean generative AI specialists.”