[AI Revolution](120) Deepfake Detection Made Easy for Everyone... SandsLab

Specialized Detection and Response Technologies for Malfunctioning 'Generative AI' Such as Deepfake and Malware

As deepfake sex crimes have recently emerged as a social issue, a domestic small and medium-sized enterprise has attracted attention by offering a service that can prevent them for free. The company is SandsLab, a security firm specializing in artificial intelligence (AI)-based cyber threat intelligence (CTI).

On the 6th, Choi Min-jun, head of AI Technology Development at SandsLab, told Asia Economy, "We are developing and researching technology in line with the national situation to prevent deepfake crimes and fake news," adding, "We hope SandsLab's deepfake detection technology will help many people as a social contribution."

The service SandsLab has launched this time is the deepfake detection solution called 'FakeCheck.' Users can directly access the website and upload images suspected of being deepfakes, and within seconds, the system analyzes and informs them of the authenticity. To make the service easily accessible to anyone, cumbersome procedures such as membership registration or authentication have been eliminated, focusing purely on functionality.

Choi explained, "FakeCheck was developed using various AI models to determine whether a person's facial image has been tampered with or forged. Since it is a service aimed at crime prevention, it does not detect deepfake images of animals or objects. However, through continuous feedback and updates, we may expand detection targets, improve detection rates, and possibly support video tampering detection in the future."

The service is provided free of charge and is available to everyone. SandsLab has decided to bear all costs related to cloud services and others required to maintain this service as part of its social contribution.

SandsLab's rapid response to deepfake crimes was largely due to its involvement in a national research and development project (IITP) related to generative AI. Currently, SandsLab is conducting research and development under the 'Core Information Security Technology Development Project' in collaboration with FortyTwoMaru, KAIST, Logpresso, and LG Uplus. SandsLab is responsible for the generative AI malfunction detection and response technology in this project.

Choi said, "The deepfake part is mainly researched by SandsLab and LG Uplus. Although only about four months have passed, so we cannot claim to have released any remarkable technology or solutions yet, we intend to perform well on the project by utilizing the technology SandsLab possesses."

SandsLab is fundamentally a company specialized in cybersecurity. One of its representative brands is 'CTX (Cyber Threat X).' CTX is based on an AI engine that analyzes files and supports queries for specific hashes (MDS, SHA1, SHA256). It provides information such as whether a file is malicious, the attack group, target countries and industries, and MITRE ATT&CK techniques.

Choi explained, "Ultimately, there comes a moment when a person directly judges 'this is a malicious file.' CTX quickly provides information about when, where, which attack group is targeting which industry and country, and so on," adding, "Because it provides persuasive information to people, it speeds up threat assessment and improves work efficiency."

SandsLab is also conducting a dataset business using CTX. A dataset refers to data collected in a related manner for a specific task. To train AI models, high-quality, structured datasets are necessary. SandsLab's datasets include information on various cyber threat attack groups and campaign-related data, helping AI algorithms respond more realistically and to diverse situations.

Choi stated, "Generative AI and dataset technologies must accompany enhanced threat detection and response capabilities," adding, "We plan to accelerate global technology development through various collaborations with domestic generative AI specialized companies and others."