Korean Vendors Target AI Guardrail Market

Korean software firms are launching products and services for the AI guardrail market, aiming to keep generative AI and AI agents within company policies and defined boundaries. AI guardrails are safeguards that limit inputs, outputs and actions according to set criteria. The term can refer narrowly to products that judge and block inputs and outputs, or more broadly to security systems across development and operations, and organization-wide governance.

As AI agents spread, controls are expanding beyond text to include actions, because an agent’s incorrect action can pose a greater security threat. Xenon, which is preparing for a KOSDAQ listing this month, recently released GenGuard as a standalone product, separating guardrail functions previously built into its generative AI platform Xenos. Users can set policies in natural language, such as “block recommendations for individual stocks,” without retraining the model.

AI security company Aim Intelligence is promoting Starfort, a real-time guardrail, alongside Stinger, an AI red-team solution. Stinger finds and assesses vulnerabilities in AI systems, while Starfort controls risks in live services. RaonSecure’s OneShield detects and blocks malicious prompt attacks such as prompt injection and jailbreaks, and automatically identifies personal information in AI inputs to anonymize or mask it.

Selectstar is developing AI reliability assessment consulting, automated red teaming and small language model-based guardrail technology. It also supports performance and safety evaluations of corporate AI models and services. Xenon Vice President and Chief Technology Officer Daewoo Myung said that as companies adopt AI at scale, security will become a key factor determining the reliability and competitiveness of AI services, beyond its role in risk management. He said the company would respond to changing security needs through its enterprise AI platform technology and industry experience, and seek to lead the AI security market suited to domestic business environments.

Korean software firms are launching products and services for the AI guardrail market, aiming to keep generative AI and AI agents within company policies and defined boundaries. AI guardrails are safeguards that limit inputs, outputs and actions according to set criteria. The term can refer narrowly to products that judge and block inputs and outputs, or more broadly to security systems across development and operations, and organization-wide governance.

As AI agents spread, controls are expanding beyond text to include actions, because an agent’s incorrect action can pose a greater security threat. Xenon, which is preparing for a KOSDAQ listing this month, recently released GenGuard as a standalone product, separating guardrail functions previously built into its generative AI platform Xenos. Users can set policies in natural language, such as “block recommendations for individual stocks,” without retraining the model.

AI security company Aim Intelligence is promoting Starfort, a real-time guardrail, alongside Stinger, an AI red-team solution. Stinger finds and assesses vulnerabilities in AI systems, while Starfort controls risks in live services. RaonSecure’s OneShield detects and blocks malicious prompt attacks such as prompt injection and jailbreaks, and automatically identifies personal information in AI inputs to anonymize or mask it.

Selectstar is developing AI reliability assessment consulting, automated red teaming and small language model-based guardrail technology. It also supports performance and safety evaluations of corporate AI models and services. Xenon Vice President and Chief Technology Officer Daewoo Myung said that as companies adopt AI at scale, security will become a key factor determining the reliability and competitiveness of AI services, beyond its role in risk management. He said the company would respond to changing security needs through its enterprise AI platform technology and industry experience, and seek to lead the AI security market suited to domestic business environments.