AI news ·
Hyundai Motor Group and AIM Intelligence adopt Nvidia's secure agent runtime as LG AI Research opens K-EXAONE-2.5 API
Hyundai is evaluating NVIDIA's agent-security tools, while LG AI Research opened API access to its K-EXAONE-2.5 model trained on 7 million synthetic Korean personas.
Hyundai Motor Group and LG AI Research are advancing Korea's AI ecosystem through separate initiatives with NVIDIA technology, focusing on secure enterprise agents and open model access. Hyundai is exploring NVIDIA NemoClaw blueprints and the OpenShell runtime for agent security, while LG AI Research has opened API access for its K-EXAONE-2.5 model trained on NVIDIA Nemotron datasets.
The security layer for autonomous agents
As AI agents handle more enterprise tasks, the infrastructure around them needs to keep sensitive data in check. NVIDIA OpenShell, part of the NVIDIA Open Agent Safety Platform, provides a secure runtime that uses sandboxed execution and policy-based controls. It governs what agents can access, which tools they can use, and how they interact with enterprise systems.
Hyundai Motor Group is evaluating how NemoClaw blueprints and OpenShell can support its internal workflows. The automaker's exploration signals growing interest in deploying agents that need guardrails before they touch production environments.
AIM Intelligence pairs its platform with OpenShell
Korean AI security company AIM Intelligence is integrating its Starfort agent control platform with NVIDIA OpenShell. The integration combines OpenShell's runtime access controls with Starfort's ability to detect malicious instructions and data leakage. Together, they create a layered defense for autonomous agent actions.
The company is considering a preliminary proof of concept for the public energy sector. Broader applications could extend into enterprise, industrial, and robotics settings where agent missteps carry operational risk.
K-EXAONE-2.5 opens for API access
On the model side, LG AI Research has opened API access for K-EXAONE-2.5. The model was developed using NVIDIA Nemotron datasets, including Nemotron-Personas-Korea - an open dataset containing 7 million fully synthetic Korean personas grounded in official Korean statistics. Model builders use it to anchor training data and evaluation in authentic Korean contexts.
NVIDIA Nemotron is a family of open models, datasets, and techniques designed to help developers build specialized, efficient AI agents. The release of K-EXAONE-2.5 API access makes a Korea-specific model available to developers who need language and cultural accuracy in their applications.
Why this matters for IT, operations, and product teams
For IT and development leaders, the Hyundai and AIM Intelligence work signals that agent security tooling is maturing beyond experimental stages. Sandboxed runtimes with policy controls are becoming a prerequisite for enterprise agent deployment, not an afterthought. Professionals evaluating AI Agent Courses should expect security architecture to feature prominently in curriculum design.
Operations and product teams in government, energy, and industrial sectors should track AIM Intelligence's proof of concept closely. If agent control platforms prove effective in high-stakes environments like public energy, adoption patterns will likely accelerate across regulated industries. For CIOs and technology strategists, the pairing of open models like K-EXAONE-2.5 with secure runtimes points toward a stack where model choice and safety infrastructure are decoupled - a pattern worth factoring into AI IT Strategy Training and procurement roadmaps.