Military AI Hinges on Who Controls Data Access

Defence planners at NATO Edge weighed how to adopt AI without surrendering sovereignty over sensitive data, a policy question still at the discussion stage.

Updated on Nov. 22, 2026 in Artificial Intelligence

Isometric editorial illustration of a steel security gate with a keyed lock in a concrete wall, a cable running through it, in muted oxblood, teal, and mustard.
NATO's 32 member states and Ukraine discussed in Izmir how to adopt military AI without surrendering control over sensitive battlefield data. AI Illustration. Upload story photo >

Live Poll

Do you trust private tech companies to be involved in national defense?

NATO Edge, a three-day conference on critical technologies and digital warfare held in Izmir, Turkey, discussed how militaries can integrate AI and high technology without giving up sovereignty over sensitive data. Military chiefs, government officials and defence experts from NATO's 32 member states and Ukraine attended. This is a policy discussion; the extraction does not describe a shipped system.

Why it matters

Militaries increasingly depend on developers who focus mainly on civilian needs, so the question of who controls access to sensitive battlefield data is becoming a central constraint on military AI. The discussion comes as Ukraine, nearly five years into its war with Russia, has faced restrictions on access to Starshield, a military satellite network owned by SpaceX.

The conference addressed integration of AI and high technology into military systems, but no specific performance benchmarks, system specifications or deployment figures were reported. The governing constraint discussed was permission, including classification boundaries, national rules and trust decisions, rather than a technical capability threshold.

The players

SpaceX

A private space and satellite company that owns Starshield, the military counterpart to its civilian Starlink satellite network.

Elon Musk

The founder and public face of SpaceX, whose company's satellite networks are central to Ukraine's access restrictions.

The details

Defence establishments are experimenting with standalone exercises that give varying levels of trust and access to outside technology firms. Companies deliver a networked battlefield by being granted some access to confidential data, which is where sovereignty over that data becomes contested. The bottleneck, as discussed, is less the technology itself than the permission structure that decides who sees what.

Timeline

  1. 2022: Russia invades Ukraine, and Kyiv's access to Starshield begins to be restricted.

  2. 2026-11-18

    NATO Edge conference discussions on military AI and data sovereignty take place in Izmir, Turkey.

  3. 2028: Target year for talks on regulating killer robots.

The Tech Race

Starshield, SpaceX's military satellite network, sits alongside the civilian Starlink constellation, and its access rules now shape how allied militaries and Ukraine can use commercial space infrastructure. Governments operating only a handful of satellites of their own depend more heavily on such commercial networks than states with large national constellations.

For most readers, the practical effect is indirect: decisions about who can access military-grade AI and satellite data shape which governments can use commercial networks in conflicts. The extraction does not specify any availability, pricing or timeline for changes to Starshield access, so those outcomes remain unannounced.

The takeaway

Watch whether the permission frameworks for military AI and satellite access, rather than the underlying technology, become the main subject of allied policy in the coming years. The 2028 target for talks on regulating killer robots is the concrete milestone to track for whether those frameworks get written down.

Further reading

For broader context on how governments are weighing AI in defence, explore our Artificial Intelligence coverage.

Live Poll

Do you trust private tech companies to be involved in national defense?