A new CSET report argues that NATO Maven should serve as an operational learning laboratory for coalition AI-enabled decision support and that lessons could inform work with Pacific partners through AUKUS Pillar II. The recommendation is grounded in a real NATO system that now has full technical capability and classified-network accreditation, but it is not an announced CDAO policy, AUKUS project, funding decision or commitment to share the system with non-NATO partners.
By Land Offset PAI
Sept. 12, 2026
WASHINGTON — The Center for Security and Emerging Technology, or CSET, has released a policy report calling on U.S. and international decision-makers to use NATO’s Maven Smart System as a learning laboratory for coalition artificial-intelligence-enabled decision support. The report recommends that the Department of War’s Chief Digital and Artificial Intelligence Office, or CDAO, draw lessons from Task Force Maven’s NATO experience to accelerate adoption among critical Pacific partners, with AUKUS Pillar II as a possible coordination forum. 1
This is a recommendation by CSET, not an announced decision by CDAO, NATO or the AUKUS partners. CSET does not say that a formal program has been approved to export, share or integrate NATO Maven with AUKUS countries or other Indo-Pacific partners. Nor does it claim that an AI-enabled decision-support system has solved the political, classification, data-sharing and interoperability problems that routinely complicate coalition operations. 1
The report’s timing is notable because NATO’s Maven Smart System, or MSS, reached full technical operational capability on June 22 and received full security accreditation to operate on NATO’s classified network. NATO says the platform connects command-and-control systems across the Alliance, supports multiple AI models and has been tested during exercises including STEADFAST DETERRENCE. 2 Those developments provide an empirical case to study. They do not automatically create a transferable coalition architecture for a different partnership, data environment or legal framework.
What CSET recommends
CSET’s September report, Coalition Operations: The Untapped Role of AI in Military Decision-Making, examines AI-enabled decision-support systems, or AI-DSS, as software platforms that ingest, analyze, visualize and share disparate data. The authors—Emelia Probasco, Sophie Mayo and Lauren Kahn—assess four coalition applications: establishing a common operating picture, supporting targeting decisions, accelerating foreign-disclosure information sharing and bridging language differences. 1
The report’s four recommendations are to build confidence through demonstrations and transparency; prepare for differing national approaches to “sovereign AI”; invest in data, data sharing and data governance; and invest in NATO Maven. 1 The final recommendation is the one highlighted in the LinkedIn post that led to this article.
| CSET recommendation | What it proposes | What it does not mean |
|---|---|---|
| Build confidence | Use demonstrations, exercises and transparent discussion of benefits and incidents to calibrate user and public confidence in AI-DSS. 1 | That exercises alone establish a system’s reliability, legal acceptability or suitability for every mission. |
| Prepare for AI sovereignty | CSET says CDAO should define technical and policy standards willing partners can use to facilitate interoperability. 1 | That partners will abandon national control, data-residency rules or procurement choices. |
| Invest in data governance | Improve allied data generation, common cybersecurity certificates, data-use agreements and cross-domain solutions. 1 | That data can be shared automatically across classifications, networks or national caveats. |
| Invest in NATO Maven | Treat Task Force Maven as an operational asset and learning laboratory, and apply lessons with Pacific partners; CSET identifies AUKUS Pillar II as a possible coordination forum. 1 | A decision to deploy NATO MSS beyond NATO, create an AUKUS Maven program or fund a specific CDAO initiative. |
The report identifies political and bureaucratic constraints—not only technical limitations—as the central obstacles. It cites incompatible networks, different security and cybersecurity standards, national restrictions on information sharing, language differences, trust in partners and trust in the technology. 1 That framing is significant: a capable analytics tool cannot independently override a classification rule, a foreign-disclosure decision, an intelligence-sharing caveat or a national command authority.
What NATO Maven is—and what it has achieved
NATO’s system is a concrete, operational capability rather than a hypothetical case study. On March 25, 2025, the NATO Communications and Information Agency finalized its acquisition of the Palantir Maven Smart System NATO for use within Allied Command Operations, according to Supreme Headquarters Allied Powers Europe, or SHAPE. 3
SHAPE describes MSS NATO as a common data-enabled warfighting capability using AI applications ranging from large language models and generative AI to machine learning. The announced applications include intelligence fusion and targeting, battlespace awareness, planning and accelerated decision-making. 3 The organization specifically cautioned that the NATO system should not be confused with the National Geospatial-Intelligence Agency’s Maven Warfighter Support System. 3
In June, SHAPE announced that MSS had met full technical operational capability and all Allied Command Operations requirements. It said the NATO Security Accreditation Board granted full security accreditation for classified-network operation, authorizing use for exercises, missions and activities. 2 SHAPE also stated that the system was tested over the preceding year, including during Exercise STEADFAST DETERRENCE 2026, and that the milestone opened the way for implementation across Allied Command Operations subordinate headquarters. 2
| Publicly documented status | Evidence | Limit of the evidence |
|---|---|---|
| Acquisition | NCIA finalized the Palantir Maven Smart System NATO acquisition for Allied Command Operations on March 25, 2025. 3 | The announcement does not disclose every contract term, data source, user population or operating rule. |
| Technical capability | SHAPE says MSS reached full technical operational capability on June 22, 2026, meeting ACO requirements. 2 | “Full technical operational capability” does not establish identical adoption, training or use at every NATO headquarters or national force. |
| Classified-network accreditation | SHAPE says the NATO Security Accreditation Board granted full accreditation for the system on NATO’s classified network. 2 | Accreditation on NATO’s network is not automatic accreditation on U.S., AUKUS or other national partner networks. |
| Exercise evidence | NATO says the platform was tested during STEADFAST DETERRENCE and other exercises. 2 | Exercise testing is not a public assessment of decision quality, targeting outcomes, legal compliance or operational effectiveness in every use case. |
| Broader implementation | SHAPE says the milestone paves the way for more use across ACO subordinate headquarters. 2 | The release does not provide a completed rollout schedule, partner-access policy or all future deployment sites. |
CSET’s recommendation is therefore not premised solely on a vendor claim or future vision. It has an actual NATO implementation example, a security-accreditation milestone and exercise experience to examine. But a learning laboratory is not the same as a finished international standard, and a lesson learned from NATO may still need substantial adaptation before it applies to Indo-Pacific coalitions.
Why transfer to Pacific partners is difficult
AUKUS Pillar II provides an existing three-nation framework for advanced capabilities. The United States, Australia and the United Kingdom have already carried out an AI-and-autonomy trial under Pillar II. In April 2023, their first joint trial deployed Australian, U.K. and U.S. AI-enabled assets in a collaborative swarm to detect and track military targets in a representative environment; the Department of War said the trial also involved live collaborative model retraining and interchange of AI models. 4
That trial demonstrates that the three governments have experience cooperating on AI and autonomy. It does not establish a joint operational decision-support system, a common classified-network environment, shared data-governance rules for all AI-DSS use cases, or an arrangement for other Pacific partners. CSET’s reference to AUKUS Pillar II is intentionally framed as a potential coordination forum, not a claim that the arrangements already exist. 1
The systems also face different operating conditions. NATO has established alliance structures, common command arrangements and a classified network environment developed over decades. AUKUS is a smaller technology-and-capability partnership centered on Australia, the United Kingdom and the United States. Critical Pacific partners outside AUKUS have varying security relationships, national rules, data policies, technical infrastructure and levels of AI maturity. Any extension of NATO lessons would require decisions about access, data sources, model governance, network accreditation, human oversight, disclosure and accountability.
NATO’s own AI strategy places these conditions at the core of responsible adoption. It calls for interoperable AI, secure data infrastructure, cyber defenses, appropriate risk or impact assessments and principles including lawfulness, accountability, explainability, reliability, governability and bias mitigation. 5 It also states that AI applications require clear human responsibility and that systems should permit appropriate human-machine interaction and disengagement when unintended behavior appears. 5
These requirements reinforce CSET’s central point. Coalition AI-DSS is not principally a problem of selecting one algorithm or platform. It is a problem of aligning policies, data, infrastructure and human decision processes among sovereign governments.
Decision support is not delegated decision authority
The report includes coalition targeting as one potential AI-DSS use case. That should not be misread as a recommendation to automate lethal decisions. CSET describes decision-support functions such as organizing complex information, generating a common operating picture, managing disclosure guidance and supporting human processes. 1 NATO’s AI principles maintain that human responsibility and accountability apply, that reliability must be tested and assured within defined use cases, and that governance mechanisms must allow intervention and deactivation. 5
A system can accelerate access to information while leaving command authority, legal review, target validation and national caveats with human officials. Whether an AI-DSS improves a decision depends on the quality and provenance of its data, the reliability of the model in its specified task, the clarity of user interfaces and the discipline of human review. Those variables are not eliminated by an interoperability agreement or a successful demonstration.
What would show the recommendation is moving into policy
The evidence to watch is concrete. An implementation decision would be visible in a CDAO, NATO or AUKUS policy statement; a defined exercise or pilot; funding; participating nations; an agreed data- and security-governance model; a network-accreditation plan; and published safeguards governing model use, human oversight and information sharing. A measurable outcome would include not merely a system demonstration but an assessment of whether authorized users can share reliable information faster and more securely without weakening national controls.
CSET’s report makes a credible case for studying NATO Maven because the system has moved from acquisition to technically operational, classified-network use. The paper does not show that coalition AI-DSS barriers have been solved, and its NATO-Maven/AUKUS connection remains a policy proposal. The defensible conclusion is therefore precise: NATO Maven is now a real allied AI decision-support case from which CDAO and Pacific partners could learn, but no public decision has yet committed them to turn that lesson into a shared AUKUS or wider Indo-Pacific system.
References
- Emelia Probasco, Sophie Mayo and Lauren Kahn, “Coalition Operations: The Untapped Role of AI in Military Decision-Making,” Center for Security and Emerging Technology, September 2026. 1
- Supreme Headquarters Allied Powers Europe, “NATO Maven Smart System Achieves Full Technical Operational Capability,” June 30, 2026. 2
- Supreme Headquarters Allied Powers Europe, “NATO Acquires AI-Enabled Warfighting System,” April 14, 2025. 3
- Department of Defense, “AUKUS Partners Demonstrate Advanced Capabilities Trial,” May 26, 2023. 4
- NATO, “Summary of the NATO Artificial Intelligence Strategy,” Oct. 22, 2021. 5