Sztuczna inteligencja to broń obosieczna. Ryzyko rośnie

Technologie podwójnego zastosowania (dual-use) coraz częściej łączą sektor cywilny i wojskowy, co zwiększa ryzyko luk w cyberbezpieczeństwie. Sztuczna inteligencja jest już wykorzystywana zarówno do obrony, jak i do ataku, a czynnik ludzki pozostaje kluczowym wyzwaniem.

– There is no military or civilian solution that is 100% certain, 100% verified. The opponent may always get to know our solution. Therefore, we must plan multilayer security so that even if one layer fails, the next ones catch the problem – said Wojciech Górecki, dual-use expert at the Cybersecurity Center of the AGH University of Science and Technology, during CYBERSEC EXPO & FORUM 2026.

Górecki emphasized the need for a 'security by design’ approach, designing solutions from the start with security and multilayer protection in mind. As an example of threats, he cited a tiny component hidden in telecommunications equipment from China, capable of transmitting data to an external system, illustrating the difficulty of fully controlling supply chains.

Czynnik ludzki wciąż najsłabszym ogniwem

The biggest challenge in cybersecurity remains the human factor and human behavior. Many hackers have primarily used sociology and psychology to breach defenses, not just advanced technology. Despite technological progress, social engineering and human elements remain critical and require constant attention.

– The military, even though it orders so much, should give more orders, like a drip for the industry, so that it can work on product development – said Górecki, pointing to the situation of local drone producers in Poland who have production potential but need stable conditions and regulations to cooperate effectively with state institutions.

AI jako broń obosieczna

Sztuczna inteligencja zmienia paradygmat cyberbezpieczeństwa. AI has multiplied data analysis capabilities thousands of times, allowing rapid detection of anomalies and vulnerabilities. On one hand, it enables automating defense, but on the other, it helps adversaries launch precise and quick attacks.

– Look what happened to the latest LLM models created in the United States. For a short time, they were available to almost the entire world, and then they were restricted. Why? It turned out they provide such enormous capabilities, both for defense and attack, that they became a strategic tool – explained Wojciech Górecki.

Górecki also noted that rapid technological changes require long-term financing for dual-use technologies. The project approach 'from point A to point B’ is insufficient, as technologies need constant development and updates in a dynamically changing cyber threat environment. He stressed the need for cooperation between ordering parties (military, police, fire brigade) and scientists and entrepreneurs to effectively define needs and select optimal solutions.

Źródło: WNP.PL, Fot. materiały prasowe CYBERSEC

Categories:

Tags: