NATO’s cyber hub already uses the models to detect vulnerabilities
OpenAI, the maker of ChatGPT, announced on November 17 that it is making its advanced cybersecurity models available to NATO, the North Atlantic Treaty Organization. The company said the models would help the military alliance protect its networks from increasingly sophisticated threats. The announcement came on the first day of the NATO Edge 26 conference in Turkey.
NATO’s Communications and Information Agency, or NCIA, was already using the models. NCIA is the alliance’s technology and cyber hub, and the organization has a team of over 3,000 civilian and military experts. OpenAI said NCIA was using AI models to probe suspicious activity, detect vulnerabilities and deploy fixes.
The partnership falls within OpenAI’s Daybreak program, through which government agencies and enterprise customers can obtain advanced models for authorized cybersecurity work. It marks OpenAI’s latest foray into cybersecurity under the program.
OpenAI said NATO had already used up to 300 billion tokens per month as part of its cybersecurity work. The company did not specify when NATO started using OpenAI’s cybersecurity models. OpenAI described tokens as the currency AI models use to process information and create responses.
Dylan Browne, general manager of NATO’s NCIA, said the models would significantly help the alliance accelerate vulnerability discovery, detect emerging threats at machine speed and adopt countermeasures.
“For NATO, these technologies strengthen our cyber resilience, enhance our operational advantage and reinforce our ability to protect critical digital infrastructure at unprecedented scale and speed,” he said.
The Wall Street Journal reported that geopolitical tensions remained elevated as NATO grappled with increasingly sophisticated threats, particularly in cyberspace. The report cited a recent cyberattack against European supermarket refrigerators and freezers that caused tens of millions of dollars in damage through spoiled food and equipment failures.
Airports and airlines across Europe also received a wave of hoax bomb threats via email in recent weeks, forcing authorities to divert flights and order emergency landings. Hundreds of schools received hoax bomb threats through the same method.
In September, OpenAI said Ukraine’s Ministry of Digital Transformation would use Daybreak to protect civilian infrastructure. The ministry’s access would include tools to identify software vulnerabilities and develop and test fixes more quickly.
The European Union Agency for Cybersecurity, also known as ENISA, used OpenAI models to identify vulnerabilities in software used across EU institutions.
The Wall Street Journal also disclosed that News Corp, the owner of Dow Jones Newswires and The Wall Street Journal, has a content-licensing partnership with OpenAI.