Purpose
By addressing the dearth of literature on the subject of cybersecurity risks and artificial intelligence (AI), this study aims to close a research gap by concentrating on the ever-changing environment of online social networks (OSNs) and technology. The main goals are to classify cyberattacks into categories like malware, phishing/spam and network intrusion detection; to identify efficient algorithms for preventing cyber threats; to review relevant literature from 2019 to 2020; and to use machine learning algorithms to detect suspicious behavior related to malware. The study offers a novel framework that suggests particular machine learning algorithms for every kind of cyber threat, hence improving cybersecurity knowledge and reaction capacities. This makes the research useful for examining the impact of cybersecurity on smart cities.
Design/methodology/approach
Thirty papers have been examined on AI and machine learning algorithms, including K-nearest-neighbor (KNN), convolutional neural networks (CNN) and Random Forest (RF), that were published in 2019 and 2020. Using analytical software (NVivo), a qualitative approach is used to retrieve pertinent data from the chosen research. The researchers divide cyberattacks into three groups: network intrusion detection, phishing/spam and malware.
Findings
The study’s conclusions center on how AI and machine learning algorithms linked to cybersecurity are reviewed in the literature, how cyberattacks are classified and how an inventive framework for identifying and reducing risks is proposed. This makes the research useful for researching the implications of cybersecurity for smart cities.
Practical implications
The practical implications of this research are noteworthy, particularly in the realms of technology, AI, machine learning and innovation. The utilization of the NVivo technique enhances decision-making in uncertain situations, making the study’s results more reliable. The findings showcase the applicability of tools in analyzing malicious cyberattacks to address issues related to social media attacks, emphasizing their practical utility. The study’s relevance is further highlighted by a real-world example, where a Kuwaiti public sector fell victim to a malware attack, underlining the importance of cybersecurity measures aligned with the New Kuwait 2035 strategic development plan. The innovative framework presented in the research guides the selection of algorithms for detecting specific malicious attacks, offering practical insights for securing information technology (IT) infrastructure in Kuwait.
Social implications
The rapid digitization in Kuwait, accelerated by the COVID-19 pandemic, underscores the pivotal role of technology in government services. Ma’murov et al. (2023) emphasize the significance of digitization, particularly in accessing and verifying COVID-19 information. The call for a dedicated digital library for preserving pandemic-related material aligns with the evolving digital landscape. Cybersecurity emerges as a critical concern in Kuwait and the Gulf Cooperation Council (GCC), necessitating transnational cooperation (Nasser Alshabib and Tiago Martins, 2022). In the local context, the inefficiency of information security systems and low awareness among government employees pose cybersecurity challenges (Abdulkareem et al., 2014). Social media’s role during the pandemic highlights its significance, yet the need for cybersecurity in this domain remains underexplored (Ma’murov et al., 2023; Safi et al., 2023).
Originality/value
The unique aspect of the paper is its in-depth investigation of the relationship between cybersecurity and AI in OSNs. It uses a special application of machine learning methods, including CNN, RF and KNN, to identify suspicious behavior patterns linked to malware. The detailed analysis of 30 research papers released between 2019 and 2020, which informs the choice of suitable algorithms for diverse cyber threats, further emphasizes the study’s uniqueness. The novel framework that has been suggested categorizes assaults and suggests certain machine learning techniques for identification, offering a useful instrument to improve comprehension and reactions to a variety of cybersecurity issues.