NATO says outdated rules are slowing the sharing of AI-generated intelligence

NATO sees governance, not data supply, as the bottleneck

NATO’s intelligence apparatus is facing a familiar modern problem: the volume and value of commercial intelligence are growing, but the rules for sharing it across the alliance have not kept pace. Speaking at the GEOINT Symposium in Aurora, Colorado, Maj. Gen. Paul Lynch, NATO’s deputy assistant secretary general for intelligence, said the alliance needs to replace outdated policies that currently force members to rely on exceptions and workarounds when passing commercially generated information between countries.

The warning is less about raw collection capacity than institutional plumbing. NATO’s 32 countries already make use of commercial data, but Lynch said the alliance needs fresh data-use policies, security classification guides, contract frameworks, and releasability rules if it wants intelligence to move quickly enough to support military decisions in a more contested security environment.

AI makes an old interoperability problem harder

Commercial intelligence is already complicated to govern. Once artificial intelligence enters the workflow, Lynch argued, the problem becomes more difficult. It is no longer only a matter of deciding who can share what. It becomes a question of which model is used, what training data shaped it, what assumptions are documented, what confidence threshold applies, and in what context the output can be trusted.

That framing is important because it suggests NATO sees AI-generated intelligence as a governance problem as much as a technical one. Models are not interchangeable black boxes. Their training provenance, limitations, and handling rules affect whether an output can be accepted across national systems. Lynch said there needs to be one common AI model and interface for use by commercial and national partners across the alliance.

Commercial providers are becoming more central

The symposium audience itself underscored the shift. Companies attending the event included vendors that track Russian military activity in the Bering Strait, monitor Chinese exercises, and helped assess damage to Iranian nuclear facilities after Operation Midnight Hammer. In other words, the commercial sector is already supplying insight relevant to live military planning and strategic awareness.

That creates pressure for NATO to formalize how such intelligence is handled. If commercial capabilities are becoming part of the alliance’s operational toolkit, ad hoc sharing arrangements are likely to look increasingly inadequate. The more time-sensitive the problem, the more costly bureaucratic friction becomes.

NATO’s old strength may become relevant again

Lynch argued that NATO has one advantage here: it already has experience building common standards. The alliance has developed hundreds of standardization agreements for areas such as air defense, maritime awareness, and data formats. His point was not that AI governance will be easy, but that NATO has an institutional model for turning many national systems into something functionally interoperable.

The urgency, in his telling, is timing. He posed the issue starkly: does NATO apply that rigor to AI before the technology outpaces the frameworks, or after? In his view, the answer will be decided within the next few years. That makes this a near-term policy contest, not a distant modernization program.

A wider alliance under pressure to move faster

The comments also landed in a political context of growing defense urgency. Lynch noted that allied spending has accelerated sharply, with European NATO members and Canada having previously reached the 2% of GDP defense target and allies later pledging to move toward 5% by 2035. The spending trend suggests governments are increasingly willing to fund hard security. The challenge now is whether institutions can modernize quickly enough to use that investment well.

The implication is straightforward. Buying sensors, commercial imagery, and AI tools is only part of the job. NATO also has to decide how outputs are classified, shared, trusted, and acted on across sovereign systems that do not naturally move at the same speed.

Lynch called the required work “unglamorous,” but that may be exactly why it matters. In intelligence alliances, advantage rarely depends only on who has data first. It also depends on whether the right people can use it fast enough. As AI-generated intelligence becomes more important, NATO is signaling that its next modernization challenge may be rules, not hardware.

This article is based on reporting by Defense News. Read the original article.