- N2SF, aimed at innovating the national security framework, requires Zero Trust principles
- SOFTCAMP supports N2SF’s mandatory requirements through identity, RBI and document security orchestration
[Datanet] The National Network Security Framework (N2SF), created to innovate Korea’s national security framework, is being put into effect. N2SF is a new security framework that classifies government networks into Classified, Sensitive and Open grades according to the importance of the work involved, and applies security control items differentially in order to satisfy both security and the promotion of data sharing.
N2SF is intended not only to improve rigid, one-size-fits-all network separation policies but also to strengthen cyber security in line with a rapidly changing digital environment. National and public institutions responsible for core state functions such as defense, diplomacy, administration and finance are also pursuing innovation through new technologies including cloud, AI and remote work.
Cyber security is something that must be considered throughout this process. Because Korea is an important target for cyber attack groups notorious around the world, a security framework at the level of national security is urgently needed.
Reflecting these demands, N2SF divides work into security grades according to the diversity and importance of the working environment, and applies security control policies suited to each grade. It promotes the convergence of new technologies through the free flow of information and creates a smarter working environment. In doing so, it guarantees the autonomy of each institution while helping them pursue safe digital innovation. Along the way, it contributes to the creation of a digital economy by developing and nurturing a range of industries including security, AI and data.
Approaches to building N2SF by information service model
In its commentary on the security control items in the National Intelligence Service’s ‘National Network Security Framework Security Guideline (Draft)’, 176 security control items are listed across six domains — authorization, authentication, separation and isolation, control, data, and information assets — along with an explanation of the priority considerations for N2SF. The security control items are to be continuously updated to reflect changes in technology.
The National Intelligence Service also published a commentary on information service models that proposes security threats and countermeasures for major information service models. That commentary suggests approaches to building an N2SF framework for a variety of service models, including the use of generative AI, the construction of cloud-based collaboration systems, and the use of new technologies on research-purpose endpoints. The plan is to continue developing and providing further models.
For the working environments attracting the greatest interest among the models used as examples in the commentary, the following approaches to building N2SF using SOFTCAMP security solutions can be proposed.

● Using generative AI in the working environment
Government and public institutions are also reviewing the use of various generative AI services to improve work efficiency, but using generative AI is not easy in the current network-separated environment. N2SF, however, defines security principles that allow government and public institutions to use generative AI, which is why it is attracting a great deal of attention.
If the use of generative AI is designed in line with the N2SF framework, an S-grade work endpoint or On-Book located in the S-grade zone connects to generative AI located in the O-grade internet zone. In that case only O-grade information may be uploaded and downloaded, and information graded S or above may not be sent to the generative AI service.
When using generative AI, it is necessary not only to control the flow of data into environments with different security levels, but also to put in place measures that prevent a variety of security threats. For example, generative AI carries a high risk of data leakage and exposure, and there is also the problem of an expanded attack surface caused by the use of unauthorized AI services. Numerous security issues can arise, including phishing and harmful sites disguised as generative AI services, the theft of AI service accounts, and vulnerabilities in generative AI itself and in its APIs.
SOFTCAMP’s ICAM and IdP solution ‘SHIELD ID’, its secure access solution ‘SHIELD Gate’, and its document security orchestration solution ‘Document Security’ can carry out these security measures.
With SOFTCAMP’s solutions in place, the user, device and location (IP) can be verified, and access can be controlled so that only authorized generative AI services can be reached. When files are sent to a generative AI service, queries and uploads can be permitted only for documents held in O-grade storage. In addition, files generated by generative AI are sent to Open (O) storage when downloaded, thereby complying with security control principles.
SOFTCAMP’s solutions also provide the ability to inspect and control the content entered as prompts when generative AI is used. They perform regular expression-based prompt inspection — checking for personal credit information, sensitive information, specific phrases and so on — and log the input text. Only text that passes inspection without problems is sent to the generative AI service, and logging is also carried out when the response is returned. Administrators can audit and manage all of a user’s generative AI inputs and outputs.

● Collaboration systems using external cloud services
Digital work in modern society requires collaboration with a variety of internal and external organizations in many different ways. The same is true for national and public institutions, and using collaboration tools delivered as SaaS is better in terms of both convenience and efficiency. Cases involving the use of collaboration tools can be divided into the following two categories.
① When connecting to an O-grade SaaS collaboration tool from an O-grade endpoint (a remote endpoint in the internet zone, a mobile device, an On-Book and so on), O-grade work information can be created and stored, and files can be exchanged with the SaaS collaboration tool.
② When connecting from an S-grade endpoint (a work endpoint on the institution’s network, or the S-grade zone within an On-Book) to an O-grade collaboration tool located in the internet zone (O-grade), the work endpoint and On-Book can create and store both S-grade and O-grade work information, but S-grade information that has been created or stored must not be sent to the collaboration tool.
Security threats that can arise in this environment include the theft of user accounts, data leakage, malware entering through SaaS, and access by unauthorized devices and users.
SOFTCAMP protects SaaS and internet access using Remote Browser Isolation (RBI) technology. When a user attempts to access SaaS or the internet, a disposable isolation container equipped with a browser is launched in an environment isolated from the endpoint. That disposable isolation container carries out the SaaS or internet access on the user’s behalf.
The disposable isolation container sends the user only what appears on screen, as a visual stream, and does not pass any script from the SaaS service or the internet to the endpoint.
Files downloaded through SaaS or the internet have malware pre-emptively removed through content disarm and reconstruction (CDR). As a result, malicious elements hidden in websites and files never reach the user’s endpoint.
In addition, data created on the endpoint is graded, and whether it may be transmitted or must be blocked is decided by grade. Data of a grade that may be transmitted is stored in an isolated file repository, after which document upload and download are controlled. Where necessary, document encryption or CDR is applied to prevent data exposure and malware infection. Finally, even after a SaaS connection is established, navigation to unauthorized or malicious sites is controlled.
● Using new technologies in R&D environments and improving development environment convenience
In an IT environment where new technologies appear almost daily, running research and development environments as closed networks does not match reality. In particular, a new environment is needed for developers who want to work freely regardless of time and place, or who need to access systems remotely because of urgent work. The N2SF guideline therefore defines policies for using new technologies in R&D environments and improving the convenience of development environments.
Because R&D endpoints can use O-grade external internet and SaaS services, security measures must also be established against threats such as the leakage of research and important data, malware infection, supply chain attacks caused by the use of malicious open source, attempts at remote desktop access and the use of unauthorized remote management tools, and access to non-work and phishing sites.
Examples of how SOFTCAMP’s solutions are used for R&D environments and for improving development environment convenience are as follows.
① Using new technologies in R&D environments
SOFTCAMP applies isolation containers so that R&D endpoints can safely access O-grade external internet, SaaS and generative AI services. Users are permitted to access only authorized sites (including SaaS and generative AI services), while access to harmful and unauthorized sites is blocked. Data leakage is controlled through clipboard control between R&D network PCs and the sites they connect to.
When files are sent to a site, only O-grade files that may be transmitted are sent, while transmission of S-grade files containing personal or sensitive information is blocked. Files downloaded from a connected site have malware blocked through file sanitization (CDR) and are protected by DRM encryption. This provides security optimized for the use of R&D endpoints, effectively preventing data leakage while keeping malware out.
② Improving development environment convenience
To make development environments more convenient, developers need to be able to access the internet and use resources such as open source, connect to endpoints remotely, and carry out their work. Under the N2SF framework, a developer can create and store both S-grade and O-grade work information by searching for and viewing internet information needed for development from a development endpoint (S-grade) located in the S-grade zone of the institution’s network.
In addition, although the internet zone is O-grade, SOFTCAMP’s solutions make it possible to work from a remote location. Access to S-grade endpoints from remote locations is granted through the reinforced identity authentication, security pledges, usage periods and access resource restrictions the solution provides.
At this point, rather than connecting directly to the S-grade endpoint from the remote location, the connection is made through an isolated environment, preventing the entry of malware. Once the isolated connection is established, clipboard use and data transfer are fully blocked, and additional security is applied through screen marking and key logging.
Users can carry out their work through the endpoint they connect to in isolation, and can access development systems (S-grade) on the institution’s network through their existing endpoint.
Protecting against third-party risk from suppliers and the supply chain
SOFTCAMP’s solutions are also proposed for mitigating third-party risk that exploits suppliers, software supply chains and the like. SHIELD Gate provides a remote access environment that guarantees VPN-level security when external partners use internal business systems. In particular, it makes it possible to achieve the same strong level of security as internally, extending even to nth-tier suppliers and customers where security policies are difficult to apply.
When a supplier user on an external network accesses an internal business system for which they have been granted permission, they are required to pass through an isolated browser and an IAP (Identity Aware Proxy) after an appropriate level of authentication for the circumstances. Downloaded files can also be viewed and edited in a web isolation repository, so collaboration can proceed safely without downloading anything to the supplier user’s PC.
Safe remote access to OT networks is also supported. When a user connects from a remote location via the web or RDP, information about the access environment is analyzed comprehensively and access is permitted only when the request is legitimate, with web isolation technology applied to strengthen security further.
Supply chain attacks caused by infected external code are also pre-emptively blocked using XSCAN, a supply chain security solution. XSCAN identifies elements in externally supplied code and containers that could give rise to malicious activity, detects other malicious and suspicious elements, and inspects the software bill of materials (SBOM) and the integrity of files. By applying only code and containers whose safety has been verified to the development environment, it lowers supply chain threats. It also uses AI to provide fast, accurate threat analysis reports.
Responding to N2SF with Zero Trust security
Because N2SF is a project that changes the national network security framework, it must be pursued with a long-term plan. Yet as new technologies emerge ever faster and new types of threats appear, establishing and carrying out a long-term plan is not easy.
The N2SF guideline was written with reference to the United States’ NIST RMF and the DoD’s Zero Trust policy. These were also referenced in Korea’s own Zero Trust guideline, which means that a security design that complies with Korea’s Zero Trust guideline can also satisfy N2SF. Because SOFTCAMP’s security solutions were designed according to Zero Trust principles, they can also comply with the N2SF guideline, which overlaps with the Zero Trust guideline.
SOFTCAMP’s solutions are equipped with a ZTCAP (Zero Trust Conditional Adaptive Policy) engine that continuously verifies trust so that work can be carried out safely. They also automatically orchestrate security policies to match on-premises, public cloud and private cloud environments, automatically converting security policies that differ by environment. In this way they help organizations maintain Zero Trust principles continuously — in today’s technology environment and in tomorrow’s — so that they can respond to new security threats and satisfy regulatory compliance requirements.
※ Read the original: Datanet — ‘N2SF’ is implemented with Zero Trust principles