The North Atlantic Treaty Organization (NATO), the military alliance of 30 countries that border the North Atlantic Ocean, this week announced that it would adopt an 18-point AI strategy and launch a “future-proofing” fund with the goal of investing around $1 billion. Military.com reports that U.S. Defense Secretary Lloyd Austin will join other NATO members in Brussels, Belgium, the alliance’s headquarters, to formally approve the plans over two days of talks.
Speaking at a news conference, Secretary-General Jens Stoltenberg said that the effort was in response to “authoritarian regimes racing to develop new technologies.” NATO’s AI strategy will cover areas including data analysis, imagery, cyberdefense, he added.
NATO said in a July press release that it was “currently finalizing” its strategy on AI” and that principles of responsible use of AI in defense will be “at the core” of the strategy. Speaking to Politico in March, NATO assistant secretary general for emerging security challenges David van Weel said that the strategy would identify ways to operate AI systems ethically, pinpoint military applications for the technology, and provide a “platform for allies to test their AI to see whether it’s up to NATO standards.” van Weel said.
“Future conflicts will be fought not just with bullets and bombs, but also with bytes and big data,” Stoltenberg said. “We must keep our technological edge.”
In a document published Friday summarizing its AI strategy, NATO emphasized the need for “collaboration and cooperation” among members on “anymatters relating to AI for transatlantic defence and security.” The document also lists the organization’s principles for “responsible use for AI,” which NATO says were developed based on members’ approaches and “relevant work in applicable international fora”:
- Lawfulness: AI applications will be developed and used in accordance with national and international law, including international humanitarian law and human rights law, as applicable.
- Responsibility and Accountability: AI applications will be developed and used with appropriate levels of judgment and care; clear human responsibility shall apply in order to ensure accountability.
- Explainability and Traceability: AI applications will be appropriately understandable and transparent, including through the use of review methodologies, sources, and procedures. This includes verification, assessment and validation mechanisms at either a NATO and/ornational level.
- Reliability: AI applications will have explicit, well-defined use cases. The safety, security, and robustness of such capabilities will be subject to testing and assurance within those use cases across their entire life cycle, including through established NATO and/ornationalcertification procedures.
- Governability: AI applications will be developed and used according to their intended functions and will allow for: appropriate human-machine interaction; the ability to detect and avoid unintended consequences; and the ability to take steps, such as disengagement or deactivation of systems, when such systems demonstrate unintended behaviour.
- Bias Mitigation: Proactive steps will be taken to minimise any unintended bias in the development and use of AI applications and in data sets.
“Underpinning the safe and responsible use of AI, NATO and allies will consciously put bias mitigation efforts into practice. This will seek to minimize those biases against individual traits, such as gender, ethnicity, or personal attributes,” the document reads. “NATO will conduct appropriate risk [and] impact assessments prior to deploying AI capabilities … NATO and allies will also conduct regular high-level dialogues, engaging technology companies at a strategic political level to be informed and help shape the development of AI-fielded technologies, creating a common understanding of the opportunities and risks arising from AI.
NATO’s overtures come after a senior cybersecurity official at the Pentagon resigned in protest because of the slow pace of technological development at the department. Speaking to the press last week, Nicolas Chaillan, former chief software officer at the Air Force, said that the U.S. has “no competing fighting chance against China” in 15 to 20 years, characterizing the AI and cyber defenses in some government agencies as being at “kindergarten level.”
In 2020, the U.S. Department of Defense (DoD) launched the AI Partnership for Defense, which consists of 13 countries from Europe and Asia to collaborate on AI use in the military context. More recently, the department announced that it plans to invest $874 million next year in AI-related technologies as a part of the army’s $2.3 billion science and technology research budget.
Much of the DoD’s spending originates from the Joint Artificial Intelligence Center (JAIC) in Washington, D.C., a government organization exploring the use and applications of AI in combat. (In news related to today’s NATO announcement, JAIC is expected to finalize its AI ethics guidelines by the end of this month.) According to an analysis by Deltek, the DoD set aside $550 million of AI obligations awarded to the top ten contractors and defense accounted for 37% of total AI spending by the U.S. government, with contractors receiving the windfall.
While U.S. — and now NATO — officials grow more vocal about China’s supposed dominance in military and defense AI, research suggests that their claims somewhat exaggerate the threat. A 2019 report from the Center for Security and Emerging Technology (CSET) shows that China is likely spending far less on AI than previously assumed, between $2 billion and $8 billion. That’s as opposed to the $70 billion figure originally shared in a speech by a top US Air Force general in 2018.
While Baidu, Tencent, SenseTime, Alibaba, and iFlytek, and some of China’s other largest companies collaborate with the government to develop AI for national defense, MIT Tech Review points out that Western nations’ attitudes could ultimately hurt U.S. AI development by focusing too much on military AI and too little on fundamental research. A recent OneZero report highlighted the way that the Pentagon uses adversaries’ reported progress to scare tech companies into working with the military, framing government contracting as an ideological choice to support the U.S. in a battle against China, Russia, and other competing states.
Speaking at the Center for Strategic and International Studies Global Security Forum in January 2020, secretary of defense Mike Esper said that DoD partnerships with the private sector are vital to the Pentagon’s aim to remain a leader in emerging technologies like AI. Among others, former Google CEO Eric Schmidt — a member of the DoD’s Defense Innovation Board — has urged lawmakers to bolster funding in the AI space while incentivizing public-private partnerships to develop AI applications across government agencies, including military agencies.
Contractors have benefited enormously from the push — Lockheed Martin alone netted $106 million in 2020 for an AI-powered “cyber radar” initiative. Tech companies including Concur, Microsoft, and Dell have contracts with U.S. Immigration and Customs Enforcement, with Microsoft pledging — then abandoning in the face of protests — to build versions of its HoloLens headsets for the U.S. Army. (Microsoft this month agreed to commission an independent human rights review of some of its deals with government agencies and law enforcement.)
Amazon and Microsoft fiercely competed for — and launched a legal battle over — the DoD’s $10 billion Joint Enterprise Defense Infrastructure (JEDI) contract, which was canceled in July after the Pentagon launched a new multivendor project. Machine learning, computer vision, facial recognition vendors including TrueFace, Clearview AI, TwoSense, and AI.Reverie also have contracts with various U.S. army branches.
For some AI and data analytics companies, like Oculus cofounder Palmer Luckey’s Anduril and Palantir, military contracts have become a top source of revenue. In October, Palantir won most of an $823 million contract to provide data and big analytics software to the U.S. army. And in July, Anduril said that it received a contract worth up to $99 million to supply the U.S. military with drones aimed at countering hostile or unauthorized drones.
While suppliers are likely to remain in abundance, the challenge for NATO will be aligning its members on AI in defense. The U.S. and others, including France and the U.K., have developed autonomous weapons technologies, but members like Belgium and Germany have expressed concerns about the implications of the technologies.