AI-Powered Language Models and the Issue of National Security

Abstrakt

This study investigates the dual-use nature of AI-powered language models, focusing on their implications for national security. Through experimental research utilizing widely accessible AI tools such as ChatGPT and Bing Chat, the authors demonstrate how these technologies can be leveraged to generate credible false information across multiple languages, even by users lacking linguistic expertise. The research highlights the ease of access to such tools globally and identifies methods by which safeguards can be circumvented to produce misleading content. The findings reveal that while advanced AI models incorporate certain protections against explicit disinformation, older versions remain susceptible to manipulation, posing significant risks to social stability and national security. The study underscores the urgent need for thematic filters, ethical frameworks, and international cooperation to establish standards and regulatory measures that limit AI misuse, particularly during crises. Ultimately, the authors advocate for responsible management and oversight of AI technologies to mitigate their potential for harm in the context of information warfare and hybrid threats.