Presentation of the KTU Consortium’s mission ‘A secure and inclusive digital society’ at the Innovation Agency’s event ‘Innovation Breakfast: How mission-driven science and innovation programs address societal challenges’.
Technology is rapidly evolving and reshaping economies, governance, and daily life. But as innovation accelerates, so too does digital risk. Technological change is no longer an abstraction, even for a country like Lithuania. From electronic signatures to digital health records, this country relies on secure systems.
Cybersecurity is not only a technical challenge, but also a societal challenge, requiring the cooperation of scientists, business leaders, and policy makers. In Lithuania, this cooperation has been realized in concrete form, namely as a government-funded national initiative. The project, coordinated by the Lithuanian Innovation Agency, aims to strengthen the country’s electronic security and digital resilience.
Under this umbrella, universities and companies with long-standing expertise work together to transform scientific knowledge into market-ready, high-value innovations. Some of these solutions have already been tested in real-world environments, such as public agencies and critical infrastructure operators. Martinas Survilas, Director of Innovation Development at the Lithuanian Innovation Agency, explains:
“Our goal is to turn the potential of Lithuania’s science into real impact – to create solutions that protect the population, strengthen trust in digital services and help build an inclusive and innovative economy. Gone are the days of isolated research. In reality, science and business must work together to respond to complex, multi-layered threats.”
National mission: a secure and inclusive electronic society
Among the three strategic national missions launched under this program, one stands out for its relevance to the global digital environment. It is a “secure and inclusive electronic society” coordinated by Kaunas University of Technology (KTU).
The mission focuses on everyday users of public and private electronic services and aims to increase cyber resilience and reduce the risk of personal data breaches, directly contributing to Lithuania’s transformation into a secure and digitally empowered society. The total amount is over 24.1 million euros.
The KTU consortium includes Lithuania’s top universities such as Vilnius University of Technology and Mykolas Romelis University, as well as leading cybersecurity companies such as NRD Cyber Security, Elsis PRO, Transcendent Group Baltics, and Baltic Institute of Advanced Technology, as well as industry association Infobalt and the Lithuanian Center for Cybercrime Competence, Research and Education.
The mission’s research and development efforts cover a wide range of cybersecurity challenges that define today’s digital environment. The team is developing smart, adaptive, and self-learning buildings. In the financial sector, new AI-powered defense systems are being built to protect FinTech companies and their users from fraud and data breaches. Industrial safety is enhanced through threat detection sensor prototypes for critical infrastructure, while hybrid threat management systems are tailored for use in public safety, education, and business environments. Other research focuses on combating disinformation through AI models that automatically detect coordinated activity from bots and trolls, and creating intelligent platforms for automated cyber threat intelligence and real-time analysis.
AI fraud: A new kind of threat
According to Dr. Lasa Burzgiene, associate professor at the Faculty of Computer Science at Kaunas University of Technology, the advent of generative artificial intelligence (GenAI) and large-scale language models (LLM) has fundamentally changed the logic of fraud against e-government services.
“Historically, the main defenses relied on pattern-based detection. For example, automated filters and firewalls were able to recognize recurring fraud patterns, typical phrases and structures,” she explains. “But GenAI has eliminated the boundaries of that ‘pattern.’ Criminals can now use generative models to craft messages that are contextually accurate. The models know how to write without grammatical errors, use accurate terminology, and can even replicate an organization’s communication style. This means that modern phishing emails no longer resemble ‘classic scams’ and are difficult for even humans to recognize, let alone automated filters.”
She emphasized that both the scale and quality of attacks are evolving, saying, “GenAI has increased scale by allowing the automated generation of thousands of different, non-repeating malicious messages. Quality has increased because these messages are personalized, multilingual, and often based on publicly available information about the victim. As a result, traditional firewalls and spam Filters have lost their effectiveness because the detector can no longer rely on formal features of words, phrases, or structures. The main change is realism, which means that modern attacks no longer look like fraud.
Dr Bruziene warns that today’s criminals have access to a wide range of AI tools. They use models such as GPT-4, GPT-5, Claude, and open source alternatives such as Llama, Falcon, and Mistral. Additionally, it also uses darker variants such as FraudGPT, WormGPT, and GhostGPT that are specifically designed for malicious activities. “They can use Celebrities and Microsoft’s VALL-E to clone audio from just a few seconds of someone speaking. To create fake faces and videos, they use StyleGAN, Stable Diffusion, DALL-E, DeepFaceLab, and lip-sync solutions like Wav2Lip and First-Order-Motion,” she points out.
“Criminals create photo-realistic mugshots, deepfake videos, and copies of documents with carefully edited metadata. LLM generates high-quality, personalized phishing text and onboarding dialogs, TTS and voice-cloning models recreate the voices of victims and employees, and image generation tools create “liveness” videos that fool verification systems. Then, an automated AI agent handles the rest of the work: account creation. These multimodal chains can circumvent both automated trust-based verification and human verification. ”
“What’s scary is how accessible all this has become. Commercial TTS solutions like Eleven Labs and open source implementations of VALL-E provide high-quality voice clones to anyone. With Stable Spread, DeepFaceLab, and similar tools, you can quickly and easily generate photorealistic images and deepfakes. Thanks to this accessibility, 1 A human operator can create hundreds of convincing, different, yet interconnected fake profiles.” We have already seen examples of this in attempts to open fake accounts at financial institutions and crypto platforms. ”
Social engineering powered by AI
Another new frontier is adaptive AI-driven social engineering. Attackers no longer rely on static scripts and instead use LLMs that adapt in real time to the victim’s reactions.
Bots start with automated reconnaissance, scraping social media, professional directories, and leaked databases to build personalized profiles. The LLM then crafts an initial message that reflects the person’s professional tone and language within the organization. If there’s no response, the system automatically switches the channel from email to SMS or Slack and changes the tone from official to urgent. If the target balks, the AI cites actual internal policies and procedures to provide plausible reassurance.
In one typical scenario, a “coworker” writes an email using their work email, follows up on LinkedIn, and calls using a cloned voice. All of this is orchestrated by connected AI tools. Dr. Brugiene describes this as a new stage in the evolution of cybercrime, saying, “Social engineering has become scalable, intelligent, and deeply personal. Each victim experiences a unique and evolving deception aimed at exploiting their psychological and behavioral weaknesses.”
Lithuania’s Cyber Defense Leadership
Lithuania’s digital ecosystem, known for its advanced e-government architecture and centralized electronic ID (eID) system, faces unique challenges. But it also represents remarkable progress. The country has been steadily rising in international indicators, ranking 25th in the world on the Chandler Good Government Index (CGGI) and 33rd on the Government AI Readiness Index (2025).
Lithuania’s AI Strategy (2021-2030), updated in 2025, prioritizes AI-driven cyber defense, anomaly detection and resilience building. The National Cyber Security Center (NKSC) will integrate AI into threat monitoring to reduce ransomware incidents by 5x between 2023 and 2024. Cooperation with NATO, ENISA and EU partners will further strengthen Lithuania’s hybrid defense capabilities.
“We see cyber resilience not just as a technical challenge, but as a foundation for democracy and economic growth,” Servilas said. “Through our mission for a secure and inclusive e-society, we are not only protecting our digital infrastructure, but also enabling our citizens to trust and participate in the digital world. AI is inevitably used for malicious purposes, but it can also be used to defend. The key is cross-sector collaboration and continuous education. This mission is one of the tools that will help us turn that idea into concrete projects, pilots and services for the people of Lithuania.”
Source link
