AI: Challenges in the Cyberspace [REPORT]

Inviting a group of experts from the National Center for Cyberspace Security, Polish Ministry of Defence, Ministry of Finances, Expert Center for Cybersecurity Training, Military University of Technology, War Studies University, and Leon Kozminski University, Defence24 Group organized an online seminar on AI in the context of cybersecurity.

We have also covered matters tied to the future of cyberspace conflict, or the role that AI would have in the armed conflict, as well as the AI ethics, tools and practices.

Jacek Raubo, an analyst for Defence24.pl and CyberDefence24.pl, moderating the seminar "AI and cybersecurity: Technologies, Tools, Challenges", emphasized not only the technology-tied and tools-related challenges, but also the law and strategic or operational thinking, or issues tied to ethics. "Our societies, also in the sphere of security, follow a set of moral and ethics-related values that would need to be correlated, when it comes to the legal systems, systemic approach towards the cyber-security, and AI. We cannot run away from that", Raubo said. 

Raubo wondered whether we are to face the world governed by the algorithms and whether we are ready to develop AI in the context of cybersecurity, and also asked a question on the dilemma that the AI creators would confront when it comes to its role in technology, ethics, defence domain, and in the private space. Raubo noted that "owning solutions today, based on broadly understood AI, is one of the defining elements of the arms race, or, simply speaking, the growth of the state".

Cyberdefence or Cyberoffense?

Whether Poland may serve as a role model in the systemic understanding of AI-driven cybersecurity was discussed by Aneta Trojanowska, Director at the Cybersecurity Department of the Polish Ministry of Defence, and Plenipotentiary for the Polish Ministry of Defence for the Cyberspace Security Matters.

"In the contemporary world, cyberspace is one of the most important areas of functioning of the state. This is an area where AI also needs to be applied, for defensive purposes.One shall note that establishment and implementation of novel, often breakthrough technologies is one of the key tasks that the MoD needs to handle", Aneta Trojanowska said. "The best example of approach as such is the implementation of the Cyber.mil.pl programme in the national defence sector, based on three pillars, including consolidation of resources through the establishment of the National Centre for Cyberspace Security, and Cyberspace Defence Forces", she added.

She emphasized that the main advantages of the above stem from the consolidation of the resources that so far were scattered at the MoD. More benefits are brought thanks to employment and retention of top-class experts at the MoD, with unique know-how and capacity, and the ability to create national ICT systems, tools, and technologies, all allowing for attaining autonomy and independence from the commonly operated solutions, thus heightening the level of national security.

The Cyber.mil.pl programme also created further benefits for the MoD and cyberspace security, promoting a broader expert and institutional cooperation, also in the international dimension.

"Establishment of specialist education and training centres had a major relevance here - such as the Expert Center for Cyber-security Training, or the Cyber.mil at class programme, along with the implementation of research and development projects, involving public and private sector entities", Trojanowska said.

Aneta Trojanowska also referred to the use of AI in cyberspace operations. "It is worth adding that the relevance of AI is scrutinized and taken into the account also outside the MoD. The Council of Ministers also adopted an AI development policy in multiple areas", said Trojanowska, mentioning legal, technological, organizational, and ethical solutions.

"The contemporary conflict is no longer the one that we used to know in the not-so-distant past. The key difference stems from their hybrid profile, and their beginnings are rooted in multi-level, covert activities, also aimed at disinformation or reconnaissance, taking place in the broadly understood cyberspace. For that reason, the development of the capabilities that the Armed Forces have to work in this \[cyber\] domain of operations shall be viewed as a priority, from the point of view of our state, and from the point of view of delivering on our allied commitments", Trojanowska summarized.

The Future of the Cyber Conflicts Who is to Win the Arms Race?

Col. Mariusz Chmielewski, PhD, Eng., Deputy Director at the National Center for Cyberspace Security, provided an answer to the question of whether the leaders in the development of AI technology would win the arms race.

"This question is being asked at the strategy level in many countries. Some states base their security strategy placing a priority emphasis on technologies. My opinion is similar here, one should not forget, however, that the strategic dimension needs to fuse the three aspects: technology, human factor, and the intellectual dimension. The AI is focused on the technological dimension, and brings in incredible automation and autonomy support capacity to numerous combat platforms and systems, acting as a multiplier of military potential", Chmielewski said.

"I cannot imagine that countries that are conscious in their establishment of the military potential do not take into account a broad implementation of AI systems as support mechanisms for C2 processes, where the AI offers high levels of precision and reliability. The current research and testing show that AI is invaluable, and the only mechanism capable of efficient processing of huge data streams. Access control, perimeter monitoring, or IMINT systems are among the examples of applications as such. The review of research and press dealing with defence technologies points to those domains as attractive ones, for establishing information advantage, and situational awareness", he added.

Chmielewski also emphasized: "We cannot run away from establishing military capabilities, and therefore supporting the military operations, without taking into account the modern AI technologies, meaning large scale use of digital tech and software".

"It is a common view that, due to the profile of the cyber-domain, we would not observe an overt declaration of war. Military action below the threshold of conflict, or cyberspace operations are among the most frequent scenarios discussed in the media, that armies of many states are getting ready for. Especially with tensions present in the international arena - the profile of cyber operations, and their covert nature and easy-to-apply camouflage, mean that these capabilities are highly sought-after, and prefered. Often, AI support is required - when identifying nodes to be infected, or when it comes to dynamic malware design", he added.

Col. Chmielewski noted that we are not observing an arms race, but rather a race involving technological developments. "Many countries invest a lot in technologies, looking at dual-use, and applying the solutions developed in the civil marketplace that remains most effective in acquisition of capital, and where the research drive is found. The experts also track the results achieved closely, and the most promising and spectacular achievements are then adapted for military-grade solutions. Automation, the autonomy of numerous processes also acts as a force multiplier for military domain technological effect", Chmielewski said.

Cyber-security and the Financial Domain

Andrzej Telesz, Director at the IT Security Department of the IT Center at the Ministry of Finances, expressed his opinion on cybersecurity in the domain of finance.

"The Fourth Industrial Revolution has become a fact. Automation and robotization of business processes have become a standard, so it is impossible to avoid the use of modern, AI-based solutions in areas so important as state finances. Especially given the fact that our budget is interconnected with the EU budget. Taking care of continuous improvement of quality of service for our stakeholders at the Ministry of Finances, most of the activities involve the use of e-tools", he said.

 The use of AI at the Ministry of Finances also is present in analysis, or audit. "Not only can the AI algorithms assess the current status, but they are also able to forecast the phenomena in the ever-changing environment. The algorithm is a self-learning one, it can analyse and provide objective assessment, and in the future, it would also be able to indicate critical areas, from the point of view of the functioning of the state. This refers to the secure nature of state finances, and the taxpayers' safety", Telesz said.

How can AI be used for training?

Paweł Dziuba, Director at the  Expert Center for Cybersecurity Training responded to a question asked by Jacek Raubo - whether the AI is a "game-changer" in the cyber-security domain.

"In the military context, we may assume that the AI would be the reasons for the future wars to involve machines armed with adaptable, specialist algorithms, while the operational activities proper would be taking place in the cyberspace. In this upcoming reality, both in the military context, as well as in other areas of life where AI becomes a norm, a deficit of experts, at all levels, is perceivable, from strategic to the tactical level. Thus it is so important to establish know-how in this area, but also to use the fact that training processes can also be AI-supported", Dziuba said.

"The use of AI makes it possible to create educational programmes tailored to the individual variation among the trainees, also taking into account the skills and expertise required at the given position - at work, or in service. Automatic system, AI, adapts the curriculum to meet the needs of the employer, taking the capacity and individual traits of the trainee, into the account", he added.

"If we refer the AI support to the cybersecurity training, the support for experts, especially in the lab and training contexts, provided by the AI, would have an impact or more efficient, effective, and adaptive establishment of know-how among the specialists, also taking into the account the current trends and needs. A smart educational environment and access to such environment would translate into a better result of training, hence providing optimal establishment of individual and team expertise, required for completion of tasks in the cybersecurity domain. So yes, AI-based education is, indirectly, a milestone, when it comes to establishing the state resilience - it translates into heightened levels of cyberspace security", Dziuba summarized.

AI Technology Research

Col. Robert Kasprzyk, Ph.D. Eng., Deputy Dean at the Faculty of Cybernetics of the Military University of Technology discussed the main directions of research, covering the defensive use of the technology.

"The AI research, similarly as research tied to other breakthrough technologies, have one purpose, from the point of view of the Armed Forces - gain an advantage or even dominate the potential adversary. The history is continuous, future conflicts are just a matter of time - maybe ones that are undeclared, since, as my colleagues said, wars will no longer be formally declared, we would not switch from peace to war in an overtly declared manner, but we may face a state of strategic uncertainty, over a long period. Maybe we already are in a war, or a cyberspace confrontation", he noted, referring to the Ukrainian crisis.

"Actors having conflicting interest are driven to gain an advantage before the "merge". Ever since the end of WWII, the USA remains in possession of such advantage, achieving domination thanks to modern technologies. The "first wave" tech includes nuclear weapons, ICBMs, spy satellites. The "second wave" also saw the US domination, thanks to the modern technologies such as stealth, or PGMs. Currently, we are speaking of "third wave" technologies that include modern means of communications, data gathering and processing solutions that can handle major quantities of data, large scale computations, and the AI that associated with huge hopes, and certain threats", Kasprzyk said.

#YT[ videoId: efhp08Yw4Ag | description: ]

"The work on the "third wave" technologies is down to the arms race, as these are dual-use technologies", he emphasized. Kasprzyk said that the technologies developed during recent years are specific. They no longer serve as a tool for supporting information-decision processes. They have created a new space - the so-called cyberspace - where our lives can be found, thus resulting in a phenomenon referred to as avatarisation.

Computer technologies are developed at an exponential, not linear pace. This results in sudden qualitative changes that cannot be foreseen when being evolutionarily used to linear changer, spread across a longer timeline. The exponential growth of technology based on Moore's law resulted in the emergence of incredibly complex systems - and then, Murphy's law - "if anything may go wrong - it will" - indeed appears to be true. It cannot be any different, given the statistical point of view, and thus we are continuously updating the software - and this means that what's updated was faulty, vulnerable.

"The cyberspace has become a space for communications and cooperation, but also a confrontation of different actors at the levels of information and technology. A certain hope is seen in the development of AI when it comes to the handling of the complexity of cyberspace phenomena. This includes tools using the machine learning algorithms as well", he added.

AI in Communications

Col. Zbigniew Piotrowski, Ph.D. Hab. Eng., Professor at the Military University of Technology, explained the impact AI has on communications. "The recipe for success - gaining the operational advantage in communications, stems, first, from owning an efficient, advanced AI algorithm, and second, from owning an efficient and effective computational platform, and third, from possession of a huge database. These three components require us to learn how to use proper technologies, to gain the aforesaid advantage. The silicon technology computational platforms are no longer sufficient. Only countries remaining in possession of quantum technology, and real-time updated, well arranged, broad data set would be able to gain dominating advantage in this area", he stressed.

"Generating real time scenarios, for instance when it comes to communication systems defence and offence involves the use of the most efficient computational platform and a real data set. The success translates into gaining of momentary, or dominating, long-term advantage over the adversary, through effective jamming of his radio communications and ensuring continuity of own radio communications", Piotrowski explained.

Scale of Cyberattacks and Modern Solutions

Oleg Orlov, regional director at BlackBerry, discussed the scale of cyber-attacks, and the role of the individual companies within that scope, considering the prospect of dynamic change in cyberspace.

"I do not need to convince you that artificial intelligence plays a key role. Let us remember that approx. more than 300 thousand virus programs are being created every day. 25 percent of global operations centres experience 1 million alerts daily. This proves the scale of the attacks taking place in cyberspace", he said.

"There is a shortage of specialists in this field, cyberterrorism is increasing everywhere. The problem concerns not only Ukraine, countries that carry out cyber-attacks use artificial intelligence", Orlov emphasized. "The use of human resources is insufficient. We need more than 90 percent of malware prevention to be performed by artificial intelligence. In 2012, the former Chief Technology Officer of McFee formed Cylance company with the main goal of using artificial intelligence to prevent new never seen malware. Cylance is part of BlackBerry today and has proven AI EPP and EDR technology to prevent zero-day attacks with 99% efficiency."

Francois Baraer from BlackBerry demonstrated during videoconference how Cylance Protect AI version 2016 (5 years old) completely disabled new live malware used by hackers in Ukraine in 2022!     

AI: Legal and Ethical Challenges

The ethical and legal dimensions of cyber-intelligence were discussed by Aleksandra Przelagińska, PhD. Hab., representing the Kozminski University. She noted that the AI methodologies used in machine learning provide us with an ability to work in multiple sectors, the defence sector included, but the defence sector has a specific nature of its own.

"There is a legal side to this matter as well. We are beginning to set the standards: On 25th January the European Parliament was analyzing the latest regulation proposal when it comes to AI - the AI Act. The AI Act outlines the European approach towards the AI, as technology that is created as a trustworthy one",  Przelagińska clarified.

"It seems that this regulation pertains mainly to the commercial space, including a variety of engines used for purchase recommendations, or identification of people in public spaces. Some of these are to be defined as high-risk applications, that would not be prohibited, but would be subject to strong standardization and control, while some - such as identification of people based on Machine Learning - are to be prohibited", she said.

Przelagińska expects a breakthrough in the AI sphere, at least in Europe. "The United States of America watches those solutions closely, some countries remain sceptical - as we do not pave the way towards new solutions, regulating them instead. The remaining ones claim that standards as such may be indispensable for safe development, acting as space for innovation and generator for interesting ideas", Przelaginska added.

Human vs Machine Learning

As noted by Katarzyna Chałubińska-Jentkiewicz, Ph.D. Hab., professor at the War Studies University, Director at the Academic Center for Cybersecurity Policy of the War Studies University, the discussion concerns machine learning systems, such as machine learning algorithms, not the actual AI.

"We are noticing a certain capacity that the computers have, to analyze and solve problems - also ones in the area of ethics. This translates into the emergence of interesting problems when it comes to the question of what is morally acceptable, and what is not, what is permissible, and what should be prohibited", she said.

"Ethics and law, development of new technologies are issues that today need to be coherent", she stressed, saying that "regulations are a must. The analytics of cyberspace and AI development, and robotics, are beyond the scope of employers. The law cannot keep up - but it is how it needs to be", Chałubińska-Jentkiewicz said.