Milan: NATO is set to update its artificial intelligence strategy to include generative AI amid an increase in cyber-attacks on critical infrastructure and interference with government agencies. The standard, described by the organisation as a sort of quality control, is intended to clarify what is expected from industries, institutions and operational end-users across the alliance regarding the application of the technology.
Advancements in AI have rapidly made it an essential part of the defence alliance’s operations, especially in the cyber security domain to identify network vulnerabilities or monitor for anomalies in data access. Now NATO is looking to standardise processes to ensure that generative AI and other new tech also be utilized effectively and safely, according to David van Wheel, NATO’s assistant secretary general for emerging security challenges.
“The AI strategy endorsed by NATO in 2021, will be up for review next year and a new version will [eventually] be adopted, which will in part include language around generative AI,” he said during a media roundtable on November 28.
Van Wheel said that while the initial strategy primarily defined the six principles guiding the responsible use of AI, NATO has been working on more recent initiatives to operationalize these concepts. “Since February, NATO has been working on an AI intelligence certification standard aiming to translate the principles outlined in the 2021 strategy into concrete checks and balances, which is expected to be completed by the end of this year,” he said.
New forms of AI, of which ChatGPT is an example, are able to generate different types of content based on data provided by users. The more information it is given, the better the model learns and produces increasingly realistic outputs. Some of the key concerns around generative AI are that given the significant amount of data it requires, the likelihood that sensitive intelligence could be misused or leaked by malicious actors is increasing. If not effectively protected, confidential documents could be exploited to create deep-fakes or spread misleading facts.
“We will make it [the revised strategy] as public as we can, to inform the public and those operating in the field of AI of what we are expecting,” van Wheel added.
An alarming number of cyber offenses have recently been reported worldwide ranging from attacks on critical infrastructure to interference with governmental agencies. For Russia specifically, cyber warfare has been an important aspect of its sustained efforts to disrupt Ukrainian networks throughout the war.
“Cyber is still playing a big role in Ukraine, it just doesn’t get the same attention as other types of attacks,” van Wheel said. “Even now, we’re seeing ongoing Russian attacks on Ukrainian data centres and energy infrastructure, which as winter approaches, will be even more critical.”
According to a Microsoft report, over the last year, 120 countries have been victims of cyber-attacks, of which nearly half of the targets were NATO member-states. “We will need AI to defend ourselves,” van Wheel said.