As AI is a disruptive technology, it is believed that it affects nearly all aspects of international security, from diplomacy, intelligence, defense, to conflict management. That’s why more and more countries are growing interested in increasing their capabilities in artificial intelligence, in order to achieve dominance and collateral hegemony in international relations. AI would be the new nuclear power rewiring the global order, and all countries could seek to have it with its dual usage, civilian and military.

This article seeks to understand how could AI technologies affect global security, through focusing on two main applications of AI that will impact global security, which are cybersecurity and cyber threats from one side, and military applications and command and control systems from another side.

AI and cyber security:

There is a significant impact of AI on the cyber domain, this importance lies at the heart of the fact that it will affect cyber capabilities (both offensive and defensive) developed and used by state or non-state actors.

From one side, it could help developing new cyber security methodologies that predict an attack before it happens or fulfil its goal, and provide unbreakable data encryption that secure the data even it has been hacked.

From another side, AI could help launching more developed and sophisticated cyber attacks, attacks can’t be discovered, attributed, or stopped, or develop new autonomous cyber weapons, lethal computer viruses, that can make physical damage.

• AI effect on cyber security:

AI has a huge impact on the cyber security field, both offensive and defensive, as it is able to analyze millions of technical events, and identifying many different types of threats. It also able to analyze the human behavior itself and predict who is the most vulnerable to phishing for example. In addition, histories of behavior build profiles on users, assets, and networks, allowing AI to detect and respond to any deviations from the established norms[1], and  determine where exactly is the weak point in the cyber environment.

In other words, AI is trying to scrape through the systems, networks, databases, and human behaviors to find any minutest deviations or anomalies that increase the possibility of cyber threat. It gives an indicator on not only what could be used to attack systems but also what is most likely to be used to attack the systems.

Various types of cyber attacks, urgently need AI to handle.  Bot attacks for example, which are vast amount distributed and automated, sometimes also AI-based, cyber attacks, it’s better to deal with by AI systems, to determine is that bots are good or bad[2], and effectively respond to them.

So, AI is able to achieve predictive cyber security, by detecting even the minutest behaviors of malware or ransomware attacks before it happens, establishing a baseline of behavior for the endpoint through a repeated training process. If something out of the ordinary occurs, AI can flag it and take action, this would help to keep abreast with cyber criminals, automate threat detection, and respond more effectively than conventional software-driven or manual techniques[3].

• AI effect on cyber threats:

On the other side AI could increase the ability of cyber threats as criminal cyber-gangs, ideological and state-supported hackers can employ those same AI techniques to defeat defenses and avoid detection. Hackers can also use AI to break through defenses and develop mutating malware that changes its structure to avoid detection[4].

Besides increasing the ability of cyber threats, AI cyber security systems can be defeated by hacking their environment. As AI systems can help guard against cyber-attacks by analyzing massive volumes of data and events, if these data have been hacked, manipulated or fed by a false information the AI conclusion and decision will be affected and may be defeated. In this case AI systems are working properly, but it was failed because it was deceived by the false information they got.

That’s why AI has a huge impact on cyber security, both offensive and defensive, but at the end its difficult to say that AI will achieve absolute cyber security, or prevent cyber-attacks, but what’s certain is that it has a huge impact on cyber measures.

• AI cyber operations and global security:

The race between AI-based cyber attacks and AI-based cyber security measures could leverage the competition between nations in cyber space, even to achieve security or hegemony, or both of them, that could result at the end in militarization of cyberspace by AI, which raise the international cyber conflicts.

AI technologies are developed by both state and non-state actors, if this technology is widely distributed, it could fall into the wrong hands, like cyber criminals, terrorist groups, or a dictatorship who could threaten the stability of the international system, especially if they succeed in sabotaging nuclear facilities, aviation systems, satellite communications, autonomous and intelligent systems, etc.

As AI systems will continue learning from each other, they could escalate the offensive measures against each other, reaching the war scenario despite of the state’s desire, reaching at the end to a state of war. Even if the states make some limitations on the AI decisions to prevent the war scenario, will the militarization of AI in cyberspace lead to a balance of power between nations or misperception of power?

Another dilemma, if AI is used in defense against another AI, and both -consequently- learn from each other, can they one-day reach a compromise that achieve their interest which may contradict with the stakeholder interest.

AI and Military Implications:

Military Implications of AI are developing and enhancing to the degree that it will change how militaries operate, of course the AI military implications didn’t make a huge breakthrough yet, but it could happen in the very near future, as countries like USA and China are now racing the time to have the advanced military AI.

• Powerful AI and misperception of power:

USA is seeking to invest more than $6 billion in AI-related research and development projects in 2021[5], another estimates of the total Chinese national and local government spending on AI applications says it is clearly in the tens of billions of dollars, at least two Chinese regional governments have each committed to investing 100 billion yuan (~$14.7 billion USD).[6]

The problem here is this AI race could be race to the bottom, if the race between counties became on who first will have the most powerful AI, especially in military, without taking into consideration how this AI is safe and reliable in the field[7], and this could lead to misperception of the powerful AI, in this case a country could think it has the powerful AI, and start to take more decisions that rely on, but at the field the AI may fail leaving the country in a difficult situation.

As AI is still need pre-designed environment, with certain unique features and elements, specific conditions and rules that help AI to work properly, if the AI has been put in situations never before encountered, or the data inputs of the real environment differ than the theoretical one, or the data has been fabricated or invented, like having dummy soldiers and weapons to camouflage and decisive AI,  this could lead the AI system to fail, or push them to take wrong decision resulting in a catastrophe especially in case of war, like killing civilian, friendly targets, or targets shouldn’t be strike, that why safe and reliable AI is important before it is powerful enough.

Even the AI is safe and powerful, there is still another possibility to fail throw data poisoning attacks that could allow adversaries to manipulate algorithms by injecting bad data into training datasets, that could prevent military applications from operating effectively and safely[8] in the field.

• Effective AI and command and control systems:

The international peace may not be threatened through AI weapons, but most probably it will be affected through decisions taken from AI systems instead of humans.

The day that command and control systems being fully based on AI, the day the world could fall in endless war, third world war, even the AI is working effectively and safely, this is because these systems are going to take decisions that achieve nations interests - which are contradicting by default-  and will not look to the world’s safety, or the humanity’s goodness, they just look narrow to the nation interest only.

For example, what If north Korea launched a nuclear missile to a US target, how could the AI command and control system will response, is that by launching a more destructive nuclear bombs to respond to North Korea? If so, to what degree the destructive power should be, is that to the degree just to deter North Korea, or the degree to eliminate it from the earth. And if it is rational decision enough to deter not to eliminate, is there another possible scenario the take the civilian’s life into consideration and prevent killing more humans?

As built, AI systems will not take mentally wrong decision, since the data they driven to analyze and learn is logical, correct, and not biases[9], the decision will be rationally correct also, but what about decisions that depend on intuition and emotions instead of information, decisions that taken form a human pilot instead of a drone, exactly like the “Stealth” movie, decisions that need drawing abstract conclusions under conditions of uncertainty[10], situations that clouded by the fog of war, or situations that are too much complex for an algorithm to judge, is AI able to feel the correct decision in that case?

Another challenging issue could obstacle AI systems from working effectively related to systems integrations. A single country has many different systems in service, some of them legacy systems, and others modern one, if the integration between these systems and the AI systems is not that perfect-  which is still a challenge- a lot of failure could be expected in command and control systems, resulting in involvement in more conflicts, as the ability of different troops to work in hegemony will not be difficult, and the problem will be more if these troops from a coalition of different countries fighting together.

Conclusion:

Although AI has an effect on nearly almost everything in our lives, but not all sorts of AI will have the same effect, the narrow AI could have many applications, because it is simple, one task oriented, and work in specific conditions, that can guarantee its success. But the powerful AI that is able to influence events course, like wars, crimes, terrorist attacks, and take autonomous correct decisions in the field, still not ready yet in most of applications, once it is ready, it will have a huge impact on international relations and global security, yet it’s difficult to be optimistic and say it will achieve prosperity of humanity unless it became safe, reliable, and fixable AI.

Dr. Ehab Abdel Hamid Khalif is head of the Media and Technology Unit, Future center for advanced research and studies, Abu Dhabi, UAE. 

[1] -Using Artificial Intelligence in Cybersecurity, (2021), Balbix, accessed Sep1, 2021 on https://www.balbix.com/insights/artificial-intelligence-in-cybersecurity/

[2] - Columbus, Louis, Detecting & Stopping Bot Attacks With Better AI, (2020), Forbes, accessed Sep 3, 2021, on

https://www.forbes.com/sites/louiscolumbus/2020/08/28/detecting--stopping-bot-attacks-with-better-ai/?sh=322291331410

[3] - Belani, Gaurav, The Use of Artificial Intelligence in Cybersecurity: A Review, IEEE, “Computer dot org”,  accessed Sep 7, 2021 on https://www.computer.org/publications/tech-news/trends/the-use-of-artificial-intelligence-in-cybersecurity

[4] - Using Artificial Intelligence in Cybersecurity, (2021), Balbix, accessed Sep1, 2021 on https://www.balbix.com/insights/artificial-intelligence-in-cybersecurity/

[5] - Harper, Jon, Federal AI Spending to Top $6 Billion, (2021), National Defense Magazine, accessed Sep 12, 2021 on https://www.nationaldefensemagazine.org/articles/2021/2/10/federal-ai-spending-to-top-$6-billion

[6] - Allen, Gregory C., Understanding China's AI Strategy, (2019), Center for a New American Security, accessed Sep 15, 2021 on  https://www.cnas.org/publications/reports/understanding-chinas-ai-strategy

[7] - Scharre, Paul, "The Real Dangers of an AI Arms Race”, (May/June 2019), Foreign Affairs 98, no. 3, accessed Sep16, 2021 on https://www.foreignaffairs.com/articles/2019-04-16/killer-apps

[8] -Policy Roundtable: Artificial Intelligence and International Security, (2020), Texas National Security Review, accessed Sep 17, 2021 on https://tnsr.org/roundtable/policy-roundtable-artificial-intelligence-and-international-security/

[9] - Policy Roundtable: Artificial Intelligence and International Security, (2020), Texas National Security Review, accessed Sep 17, 2021 on https://tnsr.org/roundtable/policy-roundtable-artificial-intelligence-and-international-security/ 

[10] - Policy Roundtable: Artificial Intelligence and International Security, (2020), Texas National Security Review, accessed Sep 17, 2021 on https://tnsr.org/roundtable/policy-roundtable-artificial-intelligence-and-international-security/