# Sovereign AI Readiness Assessment for Defense and Regulated Manufacturers > Source: https://netray.co/tools/sovereign-ai-readiness-assessment > Publisher: Netray (netray.co), AI and integration engineering for ERP and manufacturing systems. > Section: Free Tools ## Short answer Score your organization across eleven dimensions of sovereign AI readiness, from data residency and model provenance to cleared personnel and air-gapped operations. ## Category on-prem-ai ## Tool **Kind:** assessment **Questions:** #### Have you classified which data your AI workloads will process? **Help:** Sovereign AI decisions cannot be made until someone has determined whether the data is CUI, export-controlled, or classified. **Options:** ##### Yes, every data source is classified and mapped to a control level **Score:** 3 ##### Major sources classified, some long-tail sources unreviewed **Score:** 2 ##### Informal understanding, nothing documented **Score:** 1 ##### No classification exercise has been done **Score:** 0 #### Can you guarantee data residency for every component of the AI stack? **Options:** ##### Yes, all inference, storage, and logging stay inside our controlled boundary **Score:** 3 ##### Yes for inference and storage, telemetry destinations still being verified **Score:** 2 ##### Mostly, but some managed services have unclear residency **Score:** 1 ##### No, several components send data to unknown regions **Score:** 0 #### How is model provenance established and recorded? **Help:** Assessors increasingly ask where model weights originated and how their integrity is verified. **Options:** ##### Verified source, recorded checksums, and a documented internal registry **Score:** 3 ##### Trusted source with informal version tracking **Score:** 2 ##### Downloaded from a public hub with no verification record **Score:** 1 ##### Unknown provenance for at least one deployed model **Score:** 0 #### Have you reviewed model licenses for export and commercial restrictions? **Options:** ##### Yes, legal review completed for every deployed model **Score:** 3 ##### Yes for primary models, secondary models pending **Score:** 2 ##### Engineering read the license summary **Score:** 1 ##### No license review has been performed **Score:** 0 #### Can the AI system operate without any internet connectivity? **Options:** ##### Yes, fully validated air-gapped operation including updates **Score:** 3 ##### Yes for inference, but updates require a connected path **Score:** 2 ##### It would degrade but partially function **Score:** 1 ##### No, it depends on external endpoints to run **Score:** 0 #### Is access to the AI system restricted to appropriately cleared or authorized personnel? **Help:** For ITAR workloads this includes verifying US person status for anyone who can reach technical data through the assistant. **Options:** ##### Yes, enforced technically and verified against personnel records **Score:** 3 ##### Yes by policy, enforcement is partly manual **Score:** 2 ##### General corporate access controls only **Score:** 1 ##### No personnel-based restriction on AI access **Score:** 0 #### Does the retrieval layer enforce document-level entitlements at query time? **Help:** An assistant indexed across everything will surface restricted content to anyone who can ask a question unless filtering happens before the model sees candidates. **Options:** ##### Yes, candidate documents are filtered by the requesting user's entitlements **Score:** 3 ##### Partly, with separate indexes per classification level **Score:** 2 ##### Only source repositories enforce permissions, not the index **Score:** 1 ##### No entitlement enforcement in retrieval **Score:** 0 #### Is every inference request logged in a tamper-evident, auditable way? **Options:** ##### Yes, full audit trail shipped to a SIEM outside the AI platform **Score:** 3 ##### Yes, logged locally with defined retention **Score:** 2 ##### Application logs exist but are not audit-grade **Score:** 1 ##### No inference-level logging **Score:** 0 #### Do you have in-country or in-boundary infrastructure capable of hosting the workload? **Options:** ##### Yes, GPU capacity already operating inside the boundary **Score:** 3 ##### Facility and power are ready, hardware procurement in progress **Score:** 2 ##### Identified a facility but no power, cooling, or hardware plan **Score:** 1 ##### No in-boundary hosting capability **Score:** 0 #### Is your AI supply chain, including hardware and software vendors, reviewed for foreign ownership or control? **Help:** Defense customers increasingly flow down FOCI and supply chain risk requirements to AI infrastructure, not just to end items. **Options:** ##### Yes, full supply chain review documented and refreshed annually **Score:** 3 ##### Primary vendors reviewed, sub-tier suppliers not assessed **Score:** 2 ##### Informal awareness, nothing documented **Score:** 1 ##### No supply chain review **Score:** 0 #### Do you have an accountable owner and governance process for AI decisions? **Options:** ##### Yes, a named owner with a cross-functional governance board that meets regularly **Score:** 3 ##### A named owner, governance handled case by case **Score:** 2 ##### Ownership is shared informally across IT and the business **Score:** 1 ##### No accountable owner for AI governance **Score:** 0 **Bands:** #### Not ready: foundational gaps **Min:** 0 **Max:** 39 **Verdict:** Your organization has significant gaps across data classification, residency, and governance that would surface immediately in a customer audit or a DCMA review. Deploying AI against controlled data in this state creates real exposure, including the risk that an inference request constitutes an unauthorized export. Foundational work has to come before any model selection. **Recommendations:** - Complete a data classification exercise for every source you intend to index, before any technical work starts. - Name a single accountable owner for AI governance with authority to approve or block deployments. - Map data residency for every component in your current AI stack, including telemetry and logging destinations. - Pause any pilot that touches export-controlled technical data until residency and access control are proven. - Engage your compliance function now so requirements shape the architecture instead of blocking it later. #### Partially ready: close the control gaps **Min:** 40 **Max:** 69 **Verdict:** You have the essential intent and several controls in place, but specific gaps would still generate findings. Typically these are retrieval-layer entitlements, audit-grade logging, or an unreviewed supply chain. These are solvable in a matter of weeks and should be closed before production traffic reaches controlled data. **Recommendations:** - Implement entitlement filtering in the retrieval pipeline so users only ever see candidate documents they are authorized for. - Ship inference audit logs to a tamper-evident SIEM outside the AI platform with retention matching your framework. - Complete legal review of every model license for export and commercial restrictions. - Document and test fully disconnected operation, including how model and package updates arrive. - Extend supply chain review to sub-tier hardware and software vendors supporting the AI stack. #### Sovereign-ready: build with confidence **Min:** 70 **Max:** 100 **Verdict:** Your data classification, residency, provenance, access control, and governance foundations are strong enough to support AI against controlled information. You can move to model selection and deployment knowing the compliance architecture will hold up under assessment. The priority now is preserving this posture as scope expands to new data sources and user populations. **Recommendations:** - Re-run this assessment whenever you add a data source, a model version, or a new user population. - Formalize your evidence package now so assessor requests are a retrieval exercise rather than a scramble. - Add red-team testing for prompt injection and data exfiltration through generated output before each major release. - Extend the same governance model to agentic workflows, where tool access creates new export and access pathways. - Publish an approved-use policy so business teams know what they may and may not put into the assistant. ## Content **Intro:** This free sovereign AI readiness assessment scores your organization across eleven dimensions of operating artificial intelligence against controlled data inside your own boundary, and it is built for defense contractors, aerospace suppliers, and regulated manufacturers subject to ITAR, CMMC, or national data sovereignty rules. It covers data classification, residency, model provenance, licensing, air-gapped operation, personnel access, retrieval entitlements, audit logging, in-boundary infrastructure, supply chain review, and governance ownership. Most organizations discover their gap is not technology but the absence of a documented data classification and an accountable owner. **Sections:** #### How the assessment works **Body:** Eleven questions score zero to three for a maximum of thirty-three points, converted to a percentage across three bands: 0-39% is not ready, 40-69% is partially ready, and 70-100% is sovereign-ready. Score the state you could evidence to an assessor today, not the state described in a policy document. The questions deliberately mix governance and technical controls because sovereign AI fails on either. A perfectly air-gapped cluster with no retrieval entitlements will leak restricted drawings to unauthorized users, and a beautifully governed program with no in-boundary infrastructure cannot deploy at all. #### What sovereign AI actually requires **Body:** Sovereignty is not simply keeping servers in your building. It means every path data can take stays inside a boundary you control and can prove to an assessor with evidence rather than assertion. That includes paths teams routinely forget, and the forgotten ones are what generate findings. We have reviewed deployments where the inference cluster was properly isolated while the observability agent shipped prompt samples to a vendor endpoint, and others where a model was air-gapped but its tokenizer downloaded configuration on first run. Enumerate every outbound connection the stack can make and justify each one before you call the deployment sovereign. **Bullets:** - Telemetry and crash reporting from AI frameworks and drivers, which frequently default to external endpoints. - Model weight provenance with recorded checksums, since assessors now ask where the artifact came from. - Retrieval-layer entitlements enforced at query time, not just permissions on the source repositories. - Personnel access verification, because for ITAR workloads a query is potentially an export to whoever is asking. #### Reading your band and sequencing the work **Body:** Sequence matters more than total score. Data classification and an accountable owner come first, because every subsequent decision depends on knowing what data is in play and who may approve deployments. Residency mapping and retrieval entitlements come second, since those are the controls most likely to generate findings. Infrastructure and model selection come last, even though they are the parts everyone wants to start with. Organizations that begin with hardware procurement consistently end up rebuilding, because the security boundary they should have designed first ends up dictating a different architecture than the one they bought. #### How Netray delivers sovereign AI **Body:** Netray builds private AI for aerospace, defense, and electronics manufacturers where compliance is the design constraint rather than a review gate. We start with data classification and a control matrix mapped to your framework, then deploy open-weight models inside your boundary with permission-aware retrieval that inherits entitlements from SyteLine, LN, M3, ServiceMax, and your PLM system. Everything is logged to your SIEM, and we produce the evidence package your assessors and prime contractors ask for. For fully disconnected sites we deliver validated air-gapped update procedures so the system stays current without ever opening an external path. **Cta:** Get a sovereign AI control matrix and deployment roadmap mapped to your ITAR, CMMC, and customer flow-down obligations. **Faq:** #### Is an AI query over export-controlled data considered an export? **Answer:** It can be. If technical data controlled under ITAR is transmitted to a service where non-US persons could access it, that is a potential unauthorized export regardless of whether a human ever reads it. The same logic applies to an internal assistant that surfaces controlled drawings to an employee who is not a US person. This is why personnel access verification and retrieval-layer entitlements matter as much as where the servers physically sit. #### Can we use commercial open-weight models in a controlled environment? **Answer:** Generally yes, and most defense suppliers do. The requirements are provenance and licensing: download from a verified source, record the checksum, store weights in a format that cannot execute code, and have legal review the license for commercial and export restrictions. Some model licenses carry acceptable-use terms that conflict with defense applications, so the license review is not a formality. Document all of it in an internal model registry for assessors. #### How do we keep an air-gapped AI system updated? **Answer:** Through a documented, controlled transfer process rather than an ad hoc one. Stage model weights, container images, and package updates on a connected system in a lower enclave, scan and verify checksums, then move them across the boundary using your approved media or data diode procedure. Test the full update path before go-live, because the most common air-gap failure is a system that runs beautifully and then cannot be patched when a vulnerability is announced. ## Search terms sovereign AI readiness, air-gapped AI deployment assessment, defense AI compliance readiness, ITAR compliant AI deployment, CMMC AI readiness assessment, data sovereignty AI checklist, controlled environment AI deployment --- Talk to an engineer: https://netray.co/contact?intent=erp_ai | Machine catalog: https://netray.co/agent/catalog.json