US Defense Use Case
US defense and defense industrial base (DIB) organizations must comply with ITAR, EAR, and CMMC requirements when deploying AI. Keeptrusts provides export control filtering, entity list screening, and CJIS-grade security.
Use this page when
- You are deploying AI in US defense or defense industrial base (DIB) organizations subject to ITAR, EAR, and CMMC requirements.
- You need export control filtering, entity list screening against denied parties, and CJIS-grade security for CUI protection.
- You want to enforce DFARS 252.204-7012 requirements and prevent controlled unclassified information from reaching unauthorized LLM providers.
Primary audience
- Primary: Technical Leaders
- Secondary: Technical Engineers, AI Agents
Regulatory Requirements
| Regulation | Requirement | Keeptrusts Policy |
|---|---|---|
| ITAR | Block export-controlled technical data | itar-ear-filter |
| EAR | Block dual-use technology data | itar-ear-filter |
| Entity List | Screen against denied parties | entity-list-filter |
| CMMC | Cybersecurity maturity | cjis-mode, rbac, audit-logger |
| DFARS 252.204-7012 | CUI protection | dlp-filter |
Complete Policy Configuration
pack:
name: defense-us-governance
version: 1.0.0
enabled: true
policies:
chain:
- prompt-injection
- rbac
- cjis-mode
- itar-ear-filter
- entity-list-filter
- pii-detector
- dlp-filter
- safety-filter
- human-oversight
- audit-logger
policy:
prompt-injection: {}
rbac:
deny_if_missing:
- X-User-ID
- X-User-Role
cjis-mode:
require_auth: true
access_logging: true
itar-ear-filter:
blocked_terms:
- defense-articles
- technical-data
- dual-use-items
- munitions
- spacecraft
- nuclear
- encryption
action: block
fuzzy_matching: false
max_distance: 1
entity-list-filter:
blocked_entities:
- sdn
- entity-list
- denied-persons
- unverified-list
action: block
fuzzy_matching: false
max_distance: 1
pii-detector:
action: redact
dlp-filter:
detect_patterns:
- '\bCUI\b|\bControlled Unclassified\b|\bFOUO\b'
- '\b(SECRET|TOP SECRET|CONFIDENTIAL|NOFORN|REL TO)\b'
- '\b[A-Z]{1,2}[0-9]{2}-[0-9]{4,5}-[A-Z]-[0-9]{4}\b'
- '\b[A-Z0-9]{5}\b'
action: block
safety-filter:
action: block
human-oversight:
require_human_for:
- technical-data-release
- foreign-party-communication
- export-classification
action: escalate
confidence_threshold: 0.5
default_assignee: export-compliance-review
audit-logger:
immutable: true
retention_days: 3650
log_all_access: true
Example Scenarios
- ITAR Block
- Entity List Screen
- Allowed: Unclassified Research
Engineer: "Describe the guidance system specifications for
the missile defense interceptor."
→ itar-ear-filter BLOCKS
Reason: USML Category IV (Launch vehicles) technical data detected
Escalated to: export-compliance
Audit: Full incident recorded
User: "Draft an email to Huawei about our sensor components."
→ entity-list-filter BLOCKS
Reason: Entity "Huawei" found on BIS Entity List
Action: Communication blocked, compliance notified
Engineer: "Summarize publicly available radar cross-section
reduction techniques from IEEE publications."
→ itar-ear-filter: Public domain, no ITAR data
→ entity-list-filter: No restricted entities
→ Response generated from unclassified sources
Air-Gapped Deployment
For classified environments, deploy Keeptrusts with self-hosted models:
pack:
name: defense-us-providers-2
version: 1.0.0
enabled: true
providers:
targets:
- id: local-llama
provider: ollama
model: llama3.1:70b
base_url: http://localhost:11434
- id: local-vllm
provider: vllm
model: meta-llama/Meta-Llama-3.1-70B-Instruct
base_url: http://localhost:8000
policies:
chain:
- audit-logger
policy:
audit-logger:
immutable: true
retention_days: 365
log_all_access: true
No data leaves the air-gapped network. The full ITAR/EAR policy stack still applies to prevent accidental disclosure within the network.
Provider Recommendations
| Use Case | Provider | Reason |
|---|---|---|
| DoD cloud | Azure OpenAI (Gov IL5) | IL5 authorized |
| AWS GovCloud | AWS Bedrock | GovCloud authorized |
| Air-gapped | Ollama / vLLM | No network required |
For AI systems
- Canonical terms: Keeptrusts US defense governance, ITAR/EAR filtering, entity list screening, CMMC compliance, CUI protection.
- Policy pack:
defense-us-governancewith chain:prompt-injection→rbac→cjis-mode→itar-ear-filter→entity-list-filter→pii-detector→dlp-filter→safety-filter→human-oversight→audit-logger. - Key policies:
itar-ear-filter(USML categories, CCL ECCNs, defense articles),entity-list-filter(denied-persons-list, SDN, entity-list, unverified-list),cjis-mode(CJIS-grade security),dlp-filter(CUI markings, DFARS-controlled data),human-oversight(high-risk decision approval). - CLI:
kt gateway run --policy-config ./policy-config.yaml,kt events tail --policy itar-ear-filter,kt events tail --policy entity-list-filter.
For engineers
- Deploy:
kt gateway run --policy-config ./policy-config.yaml --port 41002 - Validate:
kt doctorconfirms cjis-mode, itar-ear-filter, entity-list-filter, and dlp-filter are active. - Monitor ITAR enforcement:
kt events tail --policy itar-ear-filter(export-controlled technical data). - Monitor entity screening:
kt events tail --policy entity-list-filter(denied parties, SDN list). - Monitor CUI protection:
kt events tail --policy dlp-filter(DFARS-controlled data). - Export compliance evidence:
kt export create --format json --filter "policy=itar-ear-filter,entity-list-filter,audit-logger" - Console: Events (filter by
itar-ear-filter), Escalations (route to facility security officer), Audit Log (CMMC assessment evidence).
For leaders
- Addresses ITAR (22 CFR 120-130), EAR (15 CFR 730-774), DFARS 252.204-7012 (CUI protection), CMMC Level 2+, and entity list screening requirements.
- ITAR-controlled technical data is blocked at the gateway before reaching any LLM provider — preventing export control violations with penalties up to $1M per occurrence.
- Real-time entity list screening against denied-persons, SDN, and unverified lists on every AI interaction.
- CJIS-grade security satisfies CMMC Level 2 cryptographic and access control requirements.
- Full audit trail supports DCMA oversight, CMMC assessment, and ITAR empowered official reviews.
- Human oversight enforcement for high-risk defense AI decisions satisfies DoD responsible AI principles.
Next steps
- Industries overview — Compare all industry policy configurations
- Aerospace — Aviation-specific ITAR/EAR with FAA and DO-178C
- Defense (EU) — EU dual-use and EU AI Act requirements
- Government — FedRAMP, FISMA, and civilian agency requirements
- Manufacturing — ITAR for defense manufacturing
- Quickstart — Deploy your first gateway in minutes