As generative AI and non-face-to-face services become more common among users and consumers, new security issues are emerging. Security solutions that protect sensitive information and personal information using AI technology have emerged as a major concern for financial and corporate managers.

Comtrue Technology (hereafter Comtrue Tech) held the 'Generative AI Data Security & Non-Face-to-Face Identity Verification Conference' at the Korea Science and Technology Hall on the 19th. At the conference, new security solutions responding to the changing AI environment were introduced.

Recently, generative AI has developed the ability to understand and infer user intent to the level of a doctoral-level talent, and as a result, users have increased the use of generative AI in work and daily life. In this process, sensitive information and personal information are leaked, both consciously and unconsciously.

For example, there is a concern that sensitive information may be re-learned into commercial LLMs through generative AI in the process of translating emails, attachments, etc. with foreign buyers, and the leakage of company sales and public information during the HR team's resume evaluation and analysis and press release writing process were mentioned as LLM security concerns.

As concerns about leaks of sensitive corporate information and relearning of corporate secrets become visible as industry demands, ComtrueTech has developed a solution that can provide ▲prompt filtering ▲personal information masking ▲AI data input audit system for data security.

'Sphinx AI' developed by ComtrueTech can be linked with external generative AI such as ChatGPT and Claude, as well as private LLMs built by companies themselves. At this time, it provides data security in generative AI by preventing sensitive information leaks and ensuring generative AI visibility.

On the Sphinx AI platform, where commercial and private LLMs are linked, users can filter the personal information in the prompts and attachments they enter in real time and manage it with policies such as prompt blocking, pseudonymization, and confidential keyword filtering, thereby preventing the leakage of sensitive information.

In addition, it supports more productive use management of generative AI through the generative AI token management and issue data log provision functions.

Meanwhile, Comtrue Technology said, "We are an information security solution company that has been supplying information security products for over 20 years and has focused on solutions such as personal information leak prevention and data protection," and "We are currently focusing on developing AI security technology that responds to generative AI and non-face-to-face service environments."