OpenAI Set to Bring AI Into NATO’s Classified Networks After Anthropic Exit: Report

New Delhi: OpenAI is reportedly in advanced talks to deploy its artificial intelligence systems across all North Atlantic Treaty Organisation (NATO) classified networks, according to an online report. The move could mark one of the most significant steps yet in integrating cutting‑edge AI into military and intelligence operations.

Just three days ago, OpenAI published a blog post outlining its cautious approach to military contracts. The company said it believes the US military “absolutely needs strong AI models” to face growing threats, especially as adversaries increasingly integrate AI into their systems. 

However, OpenAI stressed that it had initially avoided classified deployments because its safeguards were not ready. The company insisted it would never remove key safety features to boost performance in national security work, calling that “not the correct approach.” 

Instead, OpenAI says it has been working to ensure that any classified deployment comes with strict guardrails. Its agreement with the U.S. Department of War (DoW) includes: 

- Cloud‑only deployment with a safety stack, avoiding use on edge devices that could enable autonomous lethal weapons. 

- Legal and ethical restrictions, ensuring AI cannot independently direct weapons or make high‑stakes decisions without human oversight. 

- Compliance with U.S. law, including protections for private information under the Fourth Amendment and limits on domestic surveillance. 

- Expert involvement, with cleared OpenAI engineers and safety researchers embedded to monitor usage. 

NATO Expansion 

Now, OpenAI is considering extending this architecture to NATO’s classified networks. If finalised, the deal would give allied militaries access to advanced AI tools under the same safety and oversight protocols. 

Anthropic’s Replacement and Market Shifts 

The talks also highlight a shift in the AI industry. Anthropic, another leading AI lab, had previously been in discussions with US defense officials but was sidelined after disagreements over safeguards. OpenAI’s willingness to negotiate terms that apply to all labs, including Anthropic, was seen as an effort to de‑escalate tensions between Washington and AI companies. 

Read More: After Pentagon, More Agencies Switch to OpenAI As US Government Starts Phasing Out Anthropic