{"id":10994,"date":"2025-03-31T10:00:00","date_gmt":"2025-03-31T08:00:00","guid":{"rendered":"https:\/\/haimagazine.com\/uncategorized\/the-fifth-element-in-the-confrontation-with-ai\/"},"modified":"2025-06-26T15:34:36","modified_gmt":"2025-06-26T13:34:36","slug":"the-fifth-element-in-the-confrontation-with-ai","status":"publish","type":"post","link":"https:\/\/haimagazine.com\/en\/hai-magazine-4\/the-fifth-element-in-the-confrontation-with-ai\/","title":{"rendered":"\ud83d\udd12 The fifth element in the confrontation with AI"},"content":{"rendered":"

Inez Okulska: What exactly are the Cyberspace Defense Forces? What do they do? I have a feeling that the definition in the name doesn’t really cover the whole topic. <\/strong><\/p>

Maj. Gen. Karol Molenda: <\/strong> The Cyberspace Defense Forces are legally defined as a specialized department of the armed forces authorized to conduct operations in cyberspace across the full spectrum, which means we’re responsible for active defense, not just in military systems but also in key systems from a military perspective.<\/p>

By that, I mean systems used to exchange information and conduct military operations, as well as systems without which carrying out military operations wouldn’t be effective \u2013 like critical infrastructure related to fuel flow, logistics or transport. Our relatively new group was formed in response to a declaration made during the NATO summit in Warsaw in 2016, where cyberspace was defined as the fifth domain of operations alongside land, water, air and space.<\/p>

IO: It’s almost like the fifth element!<\/strong><\/p>

GKM:<\/strong> You could say that. But what’s interesting is that it was uniquely created by humans and is modified by humans. We had made a declaration (Cyber Defense Pledge) that we will have the skills to defend this operational domain, so we had to set up a dedicated structure for this task. But \u2013 as you rightly suspected \u2013 the Cyber Defense Forces are not just “cyber”. <\/p>

IO: And not just defense?<\/strong><\/p>

GKM:<\/strong> We operate in three main areas. The first is cryptology, including cryptography and cryptanalysis \u2013 we create and break codes, trying to uphold the traditions of our cryptologists: Rejewski, Zygalski, and R\u00f3\u017cycki. We build cryptographic algorithms compliant with the principle of quantum safe (post-quantum cryptography), deal with the generation of cryptographic documents, and create encryption devices. <\/p>

The second pillar is IT \u2013 we build and maintain the main systems and telecommunication networks for the entire armed forces. We watch over a public IT system that includes over 150,000 users, but besides that, there are also non-public systems, the Polish Mission Network at the NATO Secret level, and other high-classification systems. Basically, there’s no computer, smartphone, tablet, or any other device that we haven’t bought, set up, monitored and maintained \u2013 we take care of the entire system lifecycle, including the withdrawal and destruction of old components. <\/p>

The third pillar is cybersecurity, including CSIRT MON (Computer Security Incident Response Team) as a national-level CSIRT, which deals with the coordination and handling of incidents within the MON’s area of responsibility under the National Cybersecurity System Act. We operate non-stop, 24\/7. Our soldiers and staff work in operation rooms, constantly monitoring the security of systems. <\/p>

IO: Does the fact that attackers use artificial intelligence make these attacks more severe and harder to fend off?<\/strong><\/p>

GKM:<\/strong> The most common attacks blend cyber operations with info ops that use social engineering tricks.The attacker, unable to overcome multiple layers of technical security, tries to influence the user, hoping they’ll do something wrong. However, most of our tools catch such actions at the start of the attack chain (kill chain), minimizing the chance for a user to make a mistake. <\/p>

We’re aware of the use of deepfakes and generative AI models for more believable attacks, but over the years we’ve already seen many sophisticated tools, so this is just another one. We’re alert but not scared \u2013 I don’t think it’s a big game changer that drastically increases the number of incidents. Cybersecurity is an ongoing process, but for now we feel at ease thanks to our team’s experience after many attempted attacks. <\/p>

IO: If the defense against the opponent’s AI is well cared for, what about using it on your side?<\/strong><\/p>

GKM:<\/strong> Exactly, besides those three more traditional pillars, now we’re building innovative solutions in the field of the latest technologies. It’s often said for a reason that many groundbreaking technologies, even those that were later commercialized in the civilian market, were originally developed for military needs.Even when we think about the internet \u2013 it was originally invented for the military, and then it turned the whole world upside down, not just the military sector. Therefore, we’re trying to initiate and utilize dual use technologies, as well as emerging and disruptive technologies. <\/p>

The Artificial Intelligence Implementation Center (CISI) was established with that goal in mind. It’s our internal organizational unit, whose job is to implement artificial intelligence solutions for the military and beyond. It’s actually part of our strategy \u2013 because it’s worth mentioning that we’re the only department that has developed an artificial intelligence strategy extending up to 2039. <\/p>

\"\"<\/figure>

IO: Wow, 2039? In business, they can’t agree on what the AI market will look like in 5 years, but you have a vision for the next fifteen years! <\/strong><\/p>\n\n

GKM:<\/strong> We reach far beyond the horizon [smile]. But that’s just normal practice, the army plans capacity building 15 years ahead. Theoretically, that’s how it’s defined and how budgets are assigned. In our strategy, we’ve defined several areas where we want to implement artificial intelligence in military systems and solutions. <\/p><\/div><\/div>

IO: So how do you plan to implement AI to combat this so-called fifth element?<\/strong><\/p>

GKM:<\/strong> Artificial intelligence mechanisms didn’t just pop up two years ago with the release of generative models \u2013 these are tools that were already being used in various aspects, including by us in terms of cybersecurity.<\/p>

Our military systems are a hot target for APT groups (Advanced Persistent Threats) linked to the intelligence services of unfriendly countries, including Russia. These groups differ from ordinary cybercriminal groups because they’re not just looking for quick financial gains. They get specific tasks and funding from the secret services or military, and their job is to break into a particular system and extract information from it. They don’t give up when they face challenges \u2013 they try day by day, carving through the rock drop by drop. <\/p>

Some APT groups might also conduct an operational preparation of the environment for action. They might show up in military or critical infrastructure, not getting immediate results, but ready to, for example, cut off power, cause a blackout, stop the flow of fuel or logistics at a time that benefits them. These are real threats. Our job is to be always one step ahead of them. <\/p>

Managing such a huge infrastructure, we need to be able to spot the patterns amid the noise.When dealing with systems that have over 150,000 users, it’s crucial to catch every little hint that might suggest the enemy’s presence. And that’s exactly why AI-based solutions are perfect for spotting anomalies, detecting unusual processes, and identifying potential threats. <\/p>

IO: But these are anomalies you’ve defined, in a semi-intelligent system automating human knowledge, or in a more autonomous system that finds what you didn’t even know you were looking for. Do you feel that big technology leap in your work? <\/strong><\/p>

GKM:<\/strong> There are definitely fewer false alarms now than in older devices. In cybersecurity, we already have state-of-the-art tools with built-in AI automation, and it really does help our operators, but precisely based on a collaboration between humans and artificial intelligence. For security reasons, many of our systems are separated from the Internet (what’s known as air gaps). If an operator has access to great amounts of data on anomalies, they can use that knowledge to query models that were trained in a public network and then apply that knowledge in a classified network. This speeds up the operators’ work and boosts their efficiency. <\/p>

IO: Are any elements of your processes completely based on AI, where, for example, filtering or monitoring happens without any human involvement?<\/strong><\/p>

GKM:<\/strong> At the first level in CSIRT MON, with such extensive networks, we rely on both commercial solutions and those we develop ourselves to separate the wheat from the chaff \u2013 real threats from false alarms. These are massive amounts of data; without this automation, we’d need a huge number of operators, and their perception during a high number of attacks would still be limited. For instance, because the Polish logistics center in Rzesz\u00f3w handles almost all aid for Ukraine, we have been and are under heavy attack by APT groups. In just the first year of the full-scale war in Ukraine, we received more than five times the attacks on our infrastructure compared to the preceding year. With such a complex network, the huge number of end devices and users, and five times more attacks, the adversary would have the upper hand if we didn’t automate many processes. And we aspire to be one or even two steps ahead of the enemy. <\/p>

IO: And that’s not just in the area of cybersecurity.<\/strong><\/p>

GKM:<\/strong> Exactly! In the strategy we mentioned, we pointed out specific areas where we will be \u2013 and in some cases already are \u2013 implementing artificial intelligence systems. Besides supporting cyberattack defense, faster threat identification and neutralization, or vulnerability detection in our own resources, there are plenty of uses, including ones related to physical devices or focused on people and boosting their efficiency. <\/p>

  1. Logistics optimization \u2013 they say the military starts the operation, but logistics wins it (or lack thereof loses it). It’s a massive, absolutely crucial component that includes providing meals to soldiers, ammunition, equipment, service works, supplies and transport. Modern platforms automatically report their needs, like running out of ammo or needing repairs, and these details need to be managed properly. There are solutions that connect these data and optimize processes. <\/li>\n\n
  2. Autonomous combat systems \u2013 more support for task execution through unmanned platforms, less human involvement in high-risk tasks, development of drone swarms. From the experiences of the war in Ukraine, we know that combating drones is done through jamming (electronic warfare). Thanks to AI, drones can now identify targets and eliminate them even in conditions of strong interference. Drone swarms with implemented AI can talk to each other and make decisions on their own, carrying out their tasks more precisely. <\/li>\n\n
  3. Analysis of reconnaissance and intelligence information, which means processing a lot of data from various sources and sensors. In the military, we distinguish between sensors (devices that collect information) and effectors (devices that achieve effects). For example, the F-35 jet is both a powerful weapon and a flying sensor that carries out reconnaissance tasks over land or battlefields. AI helps spot patterns, anomalies and trends in the opponent’s behavior \u2013 tracking them but also predicting their intentions. <\/li>\n\n
  4. Supporting leaders in decision-making \u2013 analyzing various action scenarios, suggesting optimal solutions, predicting recommendations and speeding up decision-making. Traditionally, a military commander makes decisions based on analyses prepared by their staff. AI can help out by drawing conclusions from a huge amount of data that the team wouldn’t be able to analyze at the same time. <\/li>\n\n
  5. Support for simulation and training \u2013 creating training materials, realistic simulations of combat conditions, virtual training environments, personalized development paths. With an army of almost 300,000 soldiers, training needs are huge, and AI can help in running simulations and war games to test different strategies and tactics. <\/li><\/ol>
    \"\"<\/figure>

    IO: So a well-organized Excel is not enough?<\/strong><\/p>

    GKM:<\/strong> Not at all! I guess no Excel could handle such a multi-tab spreadsheet. We have our own systems, like ZWSI RON, where we have all the data about logistics and equipment. We also have our own Battlefield Management System (BMS). I think we’re the only army in NATO that actually built our own system instead of using a commercial solution. Now, as part of the Artificial Intelligence Implementation Center’s activities, we want to build another module that will connect both systems and suggest a multi-criteria optimization. <\/p>

    IO: Why did you decide to build your own system when it’s such a rare practice?<\/strong><\/p>

    GKM:<\/strong> First of all, because we could [laughs]. I’m referring to our unit’s intellectual resources. DK WOC is a unique place, also unmatched in NATO. Out of 6,500 engineers and experts in communications, IT, crypto and cyber, a thousand are programmers. So we practically have our own software house. In public administration, especially in the military field, there’s nothing else like it. <\/p>

    But it’s not like the experts were just bored, so they had to come up with a project \u2013 I’ve always thought that management systems like BMS can be a “soft underbelly” because they collect huge amounts of sensitive, top-priority tactical information. If I were the opponent, I’d focus my efforts on getting into such a system.Other NATO countries use a common commercial solution, which is great for compatibility, but if such a system is attacked, we might have a bunch of these countries in trouble all at once. From experience, I know that if a solution is commercial, meaning it’s public and available on the market, it can be thoroughly analyzed by opponents for vulnerabilities and possibilities of breaching the security systems. Besides not being available to the public, our solution is also in line with the programming best practices and the principles of counterintelligence security. It has been built in such a dispersed way that there isn’t a single person in our entire team who knows the whole system. <\/p>

    IO: So, specific teams were responsible for their puzzle piece, which eventually comes together to form a cohesive picture?<\/strong><\/p>

    GKM:<\/strong> Yes, that was the idea, we even got some praise for that system [smile]. Even though it’s ours, it’s also interoperable, works on common protocols, and communicates with those commercial BMSs used in NATO \u2013 in case of a joint operation, we’re a full-fledged partner. <\/p>

    IO: That’s really reassuring news! Although, as you pointed out, there are still many other areas where you’re just starting to implement technology. What’s the biggest hurdle in implementing artificial intelligence in the military sector? The technology itself, legal regulations or organizational culture? <\/strong><\/p>

    GKM:<\/strong> They say culture eats strategy for breakfast \u2013 and it’s true. The right approach to organizational culture is fundamental. In the Cyberspace Defense Forces, we developed a totally different culture compared to how it operated for years in the military. It’s crucial to present innovations properly \u2013 just like in sports, where the one passing the ball is responsible for a good pass. In order to receive it, the other party needs to know where it’s coming from and why. That’s our job \u2013 to convince the leaders to use the available technologies, make them feel their potential, and show their real value and benefits. Sometimes things that are obvious to engineers aren’t so obvious to less technical people. <\/p>

    IO: What about the revolution in recent years, like generative artificial intelligence, large models with big capabilities but also big limitations? Are you considering using it? The possibility of hallucinations and the uncertainty about how big models work don’t really scream military sector. <\/strong><\/p>

    GKM:<\/strong> In critical applications, we just can’t afford \u2013 and not just for ethical reasons \u2013 to implement a solution that has a 70% chance of hitting the target because that missing 30% is simply unacceptable. Only when we have technology that allows us to achieve the necessary effectiveness, even in tough conditions, do we open up a huge space for its implementation. <\/p>

    However, there are areas with less risk \u2013 like training support, where generative artificial intelligence, despite its limitations, can be really handy because it offers incredible room for creating countless variants of interaction scenarios and advanced tactical games. Anyway, it’s not just about training \u2013 actually dry running different tactics can be a huge support for commanders. It’s not about relying on hints from the model, but about the ability to track the simulated variations that a person can analyze in order to make decisions based on that knowledge. <\/p>

    IO: So more internally as support in training and simulations, not operationally?<\/strong><\/p>

    GKM:<\/strong> I’ll say this: I have the privilege of working with an amazing group of open-minded people. Since 2019, when I was building this team, I’ve focused on openness and brainstorming sessions, which we organize regularly. We approach innovations with such an open mind. Just because some technologies aren’t fully mature today doesn’t mean we should totally close ourselves off to them. We can’t ignore them. I want my engineers \u2013 and we have an exceptionally high concentration of PhDs and experts per square meter in our team \u2013 to analyze possible scenarios for using new solutions and to test them in practice. It’s the only way we can stay one step ahead of the enemy. We’re well aware that our potential adversaries are pouring tons of resources \u2013 both money and brainpower \u2013 into developing similar technologies. That’s why we should follow the path taken by countries like Israel or the United States, where they don\u2019t just sit back but actively implement and test new technologies to see if they can get a leg up on future competitors. And here, the CISI center is supposed to be the “testing ground” for research.<\/p>

    IO: Let’s circle back to the intersection of science, business and the military, which is CISI.<\/strong><\/p>

    GKM:<\/strong> Many interesting solutions are being developed at several reseach institutes, but they often end up as “shelf-ware” \u2013 projects that are shelved and never implemented because they never find active recipients. If the recipient is involved in the project at every stage and cares about the implementation, the project evolves in a completely different way. Additionally, DKWOC is not just about cybersecurity \u2013 it also organizes all systems and networks. All the resources and databases from all the military systems are right here, so we don’t have to ask anyone for data \u2013 it’s just a matter of organizing these data the right way.The “fuel” for artificial intelligence is right here with us, so it\u2019s a solid base for teamwork with research groups. Our vision is to create an ecosystem where technological thinking is generated at the top institutes or research units in our country \u2013 like at the IDEAS institute \u2013 and we implement these solutions in military systems, sometimes supporting them from the development stage. We don’t want to create another unit that’s only dedicated to research, but rather use solutions developed by others. Our experts are tasked with finding a way to adapt and practically implement them. <\/p>

    IO: It’s really uplifting because a lot of research teams could develop specific systems to improve military processes, but until they’re ordered \u2013 and that’s easier said than done from a formal standpoint \u2013 they remain in a limbo, somewhere between a researcher’s imagination and the real needs or data from the military sector.<\/strong><\/p>

    GKM:<\/strong> Actually, that “last mile” is key and needs a well-built bridge between civilian institutions and our units. In many military systems, access is limited to people with the right security clearance or Polish citizenship, but that doesn\u2019t mean that people who don\u2019t meet these criteria aren\u2019t great at creating technology or can\u2019t deliver it. <\/p>

    IO: And because of this, there’s a better chance to use solutions developed in Poland.<\/strong><\/p>

    GKM:<\/strong> Exactly right! I’m totally convinced that Poles have a huge intellectual potential and we’re capable of really great things \u2013 of course with the right support. I hope we’ll see the day when we can say that we’re using solutions based on Polish ideas and Polish technological thought. My team is also keeping that in mind and, if the chance comes up, they’ll definitely want to take it \u2013 for instance, they’ll surely be testing our big Polish language models. <\/p>","protected":false},"excerpt":{"rendered":"

    Conversation with Maj. Gen. Karol Molenda, commander of the Polish Cyberspace Defense Forces <\/p>\n","protected":false},"author":5,"featured_media":9681,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"rank_math_lock_modified_date":false,"footnotes":""},"categories":[783,673,781,674,784],"tags":[],"popular":[],"difficulty-level":[36],"ppma_author":[343,656],"class_list":["post-10994","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-industry","category-hai-magazine-4","category-hai-premium","category-issue-4","category-security","difficulty-level-easy"],"acf":[],"authors":[{"term_id":343,"user_id":5,"is_guest":0,"slug":"inez-okulska","display_name":"dr Inez Okulska","avatar_url":"https:\/\/secure.gravatar.com\/avatar\/479f0f5551a6bf974825e84cfe39166b785e5cd476e583be6a22279c2c379917?s=96&d=mm&r=g","first_name":"dr Inez","last_name":"Okulska","user_url":"","job_title":"","description":"Redaktor naczelna hAI Magazine, badaczka i wsp\u00f3\u0142autorka modeli AI (StyloMetrix, PLLuM), wyk\u0142adowczyni, Top100 Woman in AI in PL"},{"term_id":656,"user_id":278,"is_guest":0,"slug":"gen-karol-molenda","display_name":"gen. Karol Molenda","avatar_url":{"url":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/03\/1080x10803-1.png","url2x":"https:\/\/haimagazine.com\/wp-content\/uploads\/2025\/03\/1080x10803-1.png"},"first_name":"gen. Karol","last_name":"Molenda","user_url":"","job_title":"","description":"\u200bGenera\u0142 dywizji. Ekspert w dziedzinie cyberbezpiecze\u0144stwa, dow\u00f3dca Komponentu Wojsk Obrony Cyberprzestrzeni. Wsp\u00f3\u0142tw\u00f3rca zaawansowanego laboratorium informatyki \u015bledczej, inicjator utworzenia Wojsk Obrony Cyberprzestrzeni."}],"_links":{"self":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/10994","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/users\/5"}],"replies":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/comments?post=10994"}],"version-history":[{"count":1,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/10994\/revisions"}],"predecessor-version":[{"id":10995,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/posts\/10994\/revisions\/10995"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media\/9681"}],"wp:attachment":[{"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/media?parent=10994"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/categories?post=10994"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/tags?post=10994"},{"taxonomy":"popular","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/popular?post=10994"},{"taxonomy":"difficulty-level","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/difficulty-level?post=10994"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/haimagazine.com\/en\/wp-json\/wp\/v2\/ppma_author?post=10994"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}