The role of artificial intelligence in combating disinformation and information security

Abstract

Abstract. In the digital age, society faces dual challenges of widespread disinformation and escalating cyber threats to information security. Disinformation campaigns, often fueled by social media, erode public trust and can undermine democratic processes. Simultaneously, cyber attacks on information systems are growing in sophistication. Artificial intelligence (AI) has emerged as a crucial tool to address both issues, offering advanced capabilities to analyze large volumes of data and detect anomalies that humans might miss [1]. AI-driven solutions are increasingly employed to combat online falsehoods. Machine learning algorithms can scan and evaluate vast streams of content at a scale unattainable for human fact-checkers, helping identify fake news, fake accounts, and coordinated propaganda in real time. Notably, generative AI technology presents a double-edged sword in this arena. On one hand, AI text and image generators lower the barrier for producing convincing fake content such as deepfake videos, intensifying the disinformation threat. On the other hand, AI also provides powerful defenses: large language models and other AI systems show promise in automated fact-checking and content verification [2]. Recent studies indicate that AI chatbots can accurately classify misleading narratives—for example, a generative AI like ChatGPT achieved over 81% accuracy in identifying conspiracy theories in experiments [2]. Overall, generative AI plays a dual role, enabling the rapid creation of synthetic false content while simultaneously offering new opportunities for detection and verification [3]. In parallel, AI has become integral to modern information security practices. Its adaptability and pattern-recognition capabilities make it well-suited for detecting and countering cyber threats [5]. Current AI applications in cybersecurity span intrusion detection systems, malware identification, fraud prevention, and protection of critical networks [5]. By analyzing network traffic and user behavior, AI can flag anomalies and potential attacks faster and more accurately than traditional methods. Notably, AI-driven intrusion detection tools have grown to dominate cybersecurity research, reflecting an ongoing emphasis on using machine learning to identify unauthorized access [5]. AI is also used proactively to anticipate threats, employing predictive analytics to identify vulnerabilities before they are exploited [5]. However, just as in the fight against disinformation, AI in cybersecurity is a double-edged sword. Attackers are weaponizing AI to develop more sophisticated exploits—such as automated spear-phishing campaigns that evade detection [5]. These dynamics create an arms race between attackers and defenders. Moreover, challenges like high computational demands, adversarial machine learning (where attackers trick AI models), and ethical concerns about autonomous decisions remain significant hurdles [5]. Researchers emphasize the need for trustworthy AI solutions, including explainable models and robust validation, to ensure AI-driven security tools are reliable and fair [5].