RaonSecure has obtained an “AI Product and Service Certification” for its AI safety guardrail solution, ‘OneShield,’ strengthening its offensive in the public sector and financial AI security markets.
On the 14th, RaonSecure announced that it received official confirmation of OneShield’s application of AI technology under the “AI Product and Service Certification System” operated by the Korea Artificial Intelligence and Software Industry Association.
The system is a procedure to verify whether AI technology has actually been applied to products or services, in accordance with the Basic Act on Artificial Intelligence. It issues certifications after technically reviewing factors such as the characteristics of AI learning and inference, and functional contribution levels.
OneShield is a solution that controls security threats that may arise during the use of generative AI and AI agents. It analyzes input and output data exchanged between users and AI in real time, allowing or blocking it according to pre-set policies.
It prevents internal corporate information or personal data from being entered into generative AI, and also responds to attempts by attackers to bypass AI policies using malicious prompts.
Specifically, it detects and blocks malicious prompt attacks such as prompt injection and jailbreaking, automatically identifies personal information input into the AI, and anonymizes or masks it. It also filters out harmful content using a model specialized for Korean.
It also provides features to control the input of sensitive information or inappropriate use of AI in accordance with security policies set by companies or institutional investors.
RaonSecure plans to expand OneShield supply, focusing on public sector and financial institutional investors, leveraging this certification acquisition. It expects that the certification can be used as a basis for determining whether AI technology has been applied during the adoption process of AI security solutions.
The company is also responding to security demand driven by the spread of AI agents. As the number of AI agents directly performing tasks increases, it anticipates growing need to control not only input and output data but also access permissions and actions.
Lee Jung-ah, CEO of RaonSecure, stated, “As we transition from using generative AI to a stage where AI agents actually perform tasks, new threats such as sensitive information leaks, malicious prompts, and policy bypasses are also expanding.” She added, “Leveraging this certification acquisition, we will expand business opportunities centered on the public sector and financial industry.”
in TECH
