Keywords

Artificial intelligence; disruptive technology; enabling technology; security risks; security threats.

Summary of artificial intelligence

Artificial intelligence (AI) is a research branch of computer science that is concerned with developing mechanisms and technological devices that can simulate human reasoning, i.e., the intelligence that is characteristic of human beings, for decision making and problem solving. It encompasses a number of methods, such as machine learning, natural language processing, speech recognition, computer vision and automated reasoning. It is not a specific technology, but a

Lieutenant Colonel at Mato Grosso Military Police in Brazil. MSc in Social Policy at Mato Grosso Federal University. 1 MSc in Policing at UCL. Chevening Scholar.

Artificial intelligence and new security threats 2

collection of interrelated technologies that very well represents the global transition from a machine-based society to an information-based society, which will consolidate on automated, data-driven and algorithmic systems [1, 2, 3].

AI is becoming an intrinsic part of our regular decision making processes. Although existing for decades, AI is being hastily driven by the unprecedented development of computing, rapidly aggregating new advancements and playing an increasingly larger role in the most diverse sectors of society, such as industry, banking, gaming, robotics, health care, detection of disasters, border control, cybersecurity, policing, urban planning, mobility, entertainment, etc. This rapid advance in AI technology is referred as AI race. Considered a big game changer, AI is likely to contribute up to 15.7 trillion US dollars to the global economy in the next decade [1, 2, 3, 4].

In an ever-changing world, private, commercial and public spheres have to readily learn how to approach AI-related issues and concerns, and how AI may play a transformative role in society, presenting a vast bulk of opportunities, risks and challenges. Thus, AI must be responsibly developed in order to enhance its capacity to improve human life [1, 2, 3].

General implications of artificial intelligence

With what seems to be endless possibilities, it is not hard to overestimate the impact of AI on different areas such as politics, ethics, economics, industry, technology, law, environment, business. However, there still is a high level of uncertainty regarding AI development; most experts agree that, although AI has developed very quickly, its biggest changes and implications are yet to come bringing simultaneously many benefits, risks and challenges [1, 2, 3].

At the same time AI is a powerful enabler which will make manufacturing widespread, it is a disruptive technology for some industries and has the potential to cause unprecedented rapid social and economic disruption. Automotive, aviation, transportation, aerospace, entertainment, data processing and security are some of the areas in which experts consider the disruptive effects of AI will be most felt. The effects in sectors as health care and food production are still Artificial intelligence and new security threats 3

controversial. Finally, service industries would be more resistant to the changes to be introduced by AI [1, 2, 3].

The increasing deployment of artificial agents combined with automation, robotics, internet of things and additive manufacturing (popularly known as 3D printing) will enable the next industrial revolution and present new challenges to all aspects of human life. Coupled with AI, automation will be applied even to higher skilled tasks, disrupting the labor market, causing economic dislocation, changing the balance of power between nations and compromising the effectiveness of economic sanctions and embargoes. Drastic disruption to the global economy and the loss of millions of jobs are highly expected: more than 40% of the current occupation in the United States may be displaced or no longer exist in the next two to three decades as a result of the development of AI and automation. Universal basic income programmes will become a necessity [1, 2, 3, 6, 7, 15, 16].

An information-based society demands attention to the renegotiation of some scenarios such as privacy, the value of information and its effects on technological advancement. AI presents new challenges to policy, regulations and laws. It will be a hard task to keep these instruments up to the pace and intensity of the changing technology, creating significant policy gaps. New legal frameworks need to be agreed once the growing use of data and algorithms will challenge the very notion of data ownership, security and privacy. The understanding about innovation and creativity, as well as the regulations about intellectual property, patent and commercial innovation will require reevaluation [1, 2, 3, 5].

Issues regarding regulatory and legal implications surrounding AI, including liability for AI decisions and the ability to appeal an AI decision still depend on further discussion. Who is responsible for an error committed by AI systems? Who is responsible if an autonomous car hit someone? What are the role and the responsibility of the developer? What about the user? All these questions are yet to be answered [1, 2, 3, 17].

Artificial intelligence and new security threats 4

Implications of artificial intelligence on cybersecurity and policing

The nefarious and criminal use of AI technologies is not unique to AI; it is a risk associated with all technological developments. AI systems will, at the same time, be useful resources to support criminal investigations and unleash new kinds of security threats. AI will have a major role both in national and domestic security strategies mainly through the use of advanced surveillance and cybersecurity [2].

AI can support criminal investigations and police operations by allowing, e.g., the replication of a crime scene evidence or an accident. The application of AI, scientific methods, and analysis techniques to the reconstruction of criminal events will play a significant role in the investigation of crime, provision of justice and crime deterrence [1, 2, 3].

The offender is a rational decision-maker. Therefore, criminal behaviour will occur when the perceived benefits outweigh perceived costs. AI will change the benefits/costs relation for existing crimes and will unleash new kinds of security threats. There are numerous ways in which offenders can exploit AI technology to commit crimes or to challenge criminal investigations. Some of these possibilities are summarised below. [1, 2, 3, 19].

1. Autonomous weapons: the greatest concern are the human-out-of-the loop weapons, which after being programmed and activated, could be used to identify and track targets, and then deliver lethal force without human mediation, input or interaction. These fully automated lethal weapons could become powerful instruments of oppression and violence [1, 2, 18].

2. Cyber sabotage: AI systems can be compromised by undetectable and dangerous defects causing real-world damage with disastrous results. Information warfare, the increased potential of malware, mainly those with small payloads, can produce devastating effects. Cyberattacks will be less detectable and more efficient. There is potential risk of interference on government decisions, elections, public opinions, etc, by spotting network opportunities for example. Deep fake images, videos and audios will be better produced, becoming very difficult to identify [1, 10].

Artificial intelligence and new security threats 5

3. Data diet vulnerability: AI system are totally dependent on the data on which they were trained, i.e., they learn and reproduce patterns in data. Adversaries may feed the system with disinformation which can get it working in unexpected and unwanted ways. Besides, data is not perfect and may inherit the prejudices of prior decision makers, reflecting widespread society’s biases. There is also the risk of automation bias, i.e., the human tendency to ascribe more credibility to outcomes and decisions produced by artificial agents without accounting for error and bias risks [1, 8, 9].

4. Filter bubbles - the personalised web: the advancements of AI systems will increase the use of targeted political and ads messaging, the hyper-segmentation of information and the fake news phenomenon, amplifying misinformation in a vicious cycle, creating a serious threat to democracy [1, 11].

5. The Big Brother even Bigger: AI will present advancements to surveillance what can be a useful tool for criminal investigation and to provide security to citizens, but, at the same time, can be used as an oppressive structure by authoritarian governments. It will be even harder for people and the press to express their opinions freely if the government decides to pursue those who present opinions contrary to their interests [1, 8].

6. The Minority Report effect: predictive policing is considered a revolutionary innovation that works with mathematical and analytical techniques to identify targets for police intervention in order to prevent and solve crimes, and identify potential offenders and victims. A vulnerability is again the data diet; if the data fed to the system is bias, it can lead to illegitimate and discriminatory police operations, diminishing trust in the police, and presenting a great risk to fair and equitable criminal justice procedures. Great risk for a nefarious feedback loop in targeting specific groups with selection bias meeting confirmation and automation bias [1, 8, 13].

7. Artificial agents in law enforcement: the growing use of AI in law enforcement, e.g., the use of intelligent traffic cameras to produce evidence, raises concerns regarding fundamental rights. The evidence produced by nonhuman automated agents can affect some citizen’s rights, as the presumption of innocence. The potential use of robot-delivered bomb, autonomous drones and weapons, are also an important concern. [1, 2, 8, 14]

Artificial intelligence and new security threats 6

8. New manufacturing capabilities will endanger jobs imposing a series of challenges: job scarcity could push people into engaging in criminal activities or towards mass migration; the latter could threaten international relations [1, 2, 3, 6, 7, 15, 16].

The KSA role on AI

AI has important geopolitical implications that must be taken into consideration. The Kingdom of Saudi Arabia (KSA) can not be out of the AI race and should considerer AI development as a strategic investment crucial to national growth, positioning the nation among those that will lead AI initiatives, becoming less dependent on importing AI systems that may not be suitable for the local context. Investment in the core science of AI, as well as in AI skills, is mandatory. Because AI is data-driven, effective digital infrastructure is essential to its successful implementation making AI secure and accessible, as well as the preparation of workers in the computer science and data science fields [1, 2, 3].

Industries, world-class leading universities, and researchers will be the main drivers of AI development, pushing AI technology to new levels, presenting constant advancements and new applications. The King Abdullah University of Science and Technology (KAUST) Visualisation Core Lab, for example, is already using data-driven and algorithm systems to understand the dynamics of an air crash.

To ensure that AI is implemented safely and sustainably, the KSA must develop the protection of privacy, human rights and fairness regarding the use of big data and automated systems, fostering public understanding and awareness of AI, improving responsive governance and regulatory mechanisms. An important initiative would be to establish an independent body to provide oversight to the development and use of AI systems. Decision-makers must recognise that AI is a potentially novel socioeconomic regime, adjusting evaluations of policy risk accordingly, distinguishing the acceptable and unacceptable uses and examining the ethical, legal and social considerations required to enable and support broad development. Partnerships between government, industry and the community will be essential in determining and developing the values underpinning AI [2].

Artificial intelligence and new security threats 7

Crime is a complex phenomenon that requires a multilateral effort to be prevented. State, citizens, and stakeholders should be involved in decision-making processes aimed at controlling crime. Identifying the potential crime preventers and promoters regarding the use of AI is important to identify who owns the responsibility to set situational crime prevention strategies. The KSA must consider the role of potential crime preventers and crime promoters to anticipate potential threats. Some are considered below [20].

Preventers

• Industry and business.

• Designers and engineers.

• International organisations and government agencies.

• Law makers.

Promoters

• Hackers.

• Terrorist groups.

• Fraudsters.

Artificial intelligence and new security threats 8

REFERENCES