OpenAI Considers Bringing ChatGPT-Style AI to NATO’s Unclassified Networks

Personnel across the North Atlantic Treaty Organisation (NATO) could soon gain access to artificial intelligence tools built on the same technology that powers ChatGPT, as OpenAI explores a potential partnership to provide AI capabilities for the alliance’s unclassified networks.

According to reporting discussions are underway about deploying OpenAI models across parts of NATO’s public IT infrastructure. The talks remain exploratory and may not ultimately result in a formal agreement.

Initial comments from OpenAI chief executive Sam Altman suggested the technology could extend across NATO’s broader systems, but a company spokesperson later clarified that Altman had misspoke and that any deployment under consideration would be limited to unclassified environments.

Even so, the discussions signal OpenAI’s growing interest in supplying AI systems to large public-sector institutions worldwide, particularly in defence and government sectors.

Pentagon backdrop

The potential NATO collaboration comes amid a shifting landscape in the U.S. defence AI market. Negotiations between the U.S.

Department of Defense and the AI firm Anthropic reportedly broke down earlier over concerns about how artificial intelligence could be used in sensitive applications, including mass surveillance and fully autonomous weapons systems.

Following the collapse of those talks, OpenAI emerged as an alternative provider for certain AI capabilities.

Altman later acknowledged that OpenAI’s agreement with the Pentagon had created reputational challenges for the company.

During an internal meeting, he reportedly described the arrangement as a “complex, but right decision” that had produced difficult public-relations consequences in the short term.

Guardrails and restrictions

OpenAI has said that its AI systems deployed for government clients will continue to operate under strict safeguards.

In a statement issued after a Pentagon agreement closed on 27 February, the company said its models would not be intentionally used for domestic surveillance of U.S. citizens or residents.

Officials also indicated that the technology would not be provided to intelligence agencies such as the National Security Agency.

The company outlined three explicit restrictions governing government deployments:

No mass domestic surveillance
No fully autonomous weapons systems
No automated high-stakes decision-making without human oversight

Under the framework, AI models are delivered through secured cloud infrastructure, with access limited to personnel who hold appropriate security clearances.

Altman has also reportedly emphasised internally that OpenAI’s role would be technical rather than operational — providing expertise on deploying AI systems rather than determining how governments ultimately use them.

NATO discussions still preliminary

For now, NATO’s potential use of OpenAI models remains under discussion. The alliance has not publicly confirmed negotiations and did not respond to requests for comment.

However, the talks reflect a broader strategic shift within OpenAI as it expands beyond consumer products such as ChatGPT into government and defence partnerships.

The company launched its OpenAI for Government program in mid-2025, initially aimed at U.S. agencies. A deal with NATO would mark one of the program’s first significant moves into Europe and multinational institutions.

European sovereignty concerns

Any wider adoption of OpenAI systems across European governments may prove more complicated.

Data sovereignty rules, local hosting requirements and concerns over strategic dependence on foreign technology companies remain major political issues across the continent.

While European policymakers are pushing for home-grown AI capability, the number of companies capable of competing at the highest level remains limited.

France’s Mistral AI is widely regarded as the region’s leading model developer, but its systems still compete with powerful offerings from global players including OpenAI, Google and Microsoft.

Whether NATO ultimately adopts OpenAI’s technology will depend on the outcome of ongoing negotiations — but the discussions highlight how rapidly artificial intelligence is becoming embedded within the infrastructure of governments, militaries and international institutions worldwide.