NIST 800-171 Controls · System and Communications Protection

SC.2.007 — Prevent Remote Device Split Tunneling

CMMC Level 2SPRS if unmet: -1MEDIUM priority~4h to implement

Prevent remote devices from simultaneously establishing non-remote connections with organizational systems and communicating via some other connection to resources in external networks (i.e., split tunneling).

What SC.2.007 means in plain English

When employees connect to your company network through VPN from home, all their internet traffic should go through the VPN — not just the traffic to company resources. Without this, an attacker on their home network could potentially use the employee's computer as a bridge into your company network. Configure your VPN to use "full tunnel" mode, not "split tunnel." This means their Netflix might be slower while on VPN, but your network stays safe.

The assessment question

“Is your VPN configured to use full-tunnel mode (not split-tunnel) so that all traffic from remote devices passes through the organizational network when connected via VPN?”

How to implement SC.2.007

Check your current VPN configuration: in pfSense OpenVPN settings, ensure "Redirect IPv4 Gateway" is checked. In WireGuard, set AllowedIPs to 0.0.0.0/0 in the client config.
If using Windows built-in VPN, go to the VPN connection properties > Networking > TCP/IPv4 Properties > Advanced, and check "Use default gateway on remote network."
Test full-tunnel is working: while connected to VPN, visit whatismyip.com — it should show the company's public IP, not the employee's home IP.
Document the VPN configuration showing full-tunnel mode is enforced and include it in your network documentation.

Evidence your assessor will ask for

VPN server configuration showing full-tunnel (not split-tunnel) is enforced
VPN client configuration files showing all traffic routes through the tunnel
Test results confirming full-tunnel is active (IP address verification while on VPN)
Written VPN policy requiring full-tunnel mode for all remote connections

Does AI prompt monitoring help with this control? Honestly, no.

SC.2.007 is met through the remediation steps above, not through AI traffic controls — an AI prompt firewall neither satisfies nor substitutes for it. We map AI monitoring only to the controls it genuinely evidences (flow control, boundary protection, audit, and incident support); for the full picture of where it does help, see the mapping guide linked below.

Full mapping: which 800-171 controls AI prompt monitoring evidences →

More System and Communications Protection controls

SC.1.001

Monitor and Protect Communications at Boundaries

SC.1.005

Implement Subnetworks for Public Components

SC.2.002

Employ Effective Security Architecture

SC.2.003

Separate User and System Management Functionality

SC.2.004

Prevent Unauthorized Info Transfer via Shared Resources

SC.2.006

Deny Network Traffic by Default

Score yourself against all 110 controls

The free ShieldReady assessment walks every NIST 800-171 requirement, computes your SPRS score, and shows exactly which gaps cost the most points.

Start the free assessment $499 AI risk assessment