The rapid advancement of synthetic media, while beneficial, has also spawned GAN-generated deepfakes, which pose risks, including misinformation and digital fraud. This paper investigates the detectability of GAN-generated static images, focusing on residual artifacts that are imperceptible to humans but detectable through digital analysis. Our approach introduces three key advancements: (1) a taxonomy for classifying GAN residues in deepfake detection; (2) a unique mixed dataset combining StyleGAN3, ProGAN, and InterfaceGAN to aid cross-model detection research; and (3) a combination of frequency space analysis and RGB color correlation methods to improve artifact detection. Covering three different transform methods, three GAN models, and twelve classification methods, ours is the most comprehensive study of detection of static deepfake face images produced by GANs. Our results demonstrate that artifact-based detection can achieve high accuracy, precision, recall, and F1 scores, challenging prior assumptions about the detectability of synthetic face images.

Given the rapid advancement of deepfake technology, which allows for the creation of highly realistic fake content, there is a pressing need for an efficient solution to address the security risks associated with this technology. Deepfake videos are widely recognized for their significant implications, including the potential for identity theft, the dissemination of false information, and the endangerment of national security. Therefore, it is crucial to develop and enhance the reliability of deepfake detection algorithms. In this study, feature extraction techniques were performed to utilize deep learning algorithms such as Xception and ResNet50 to detect deepfakes in a video dataset using the DFDC dataset. Additionally, a total of eight hybrid models were developed using various classification algorithms such as SVM, KNN, MLP, and RF. The ResNet50 and RF hybrid models achieved the highest accuracy rate of 98%, with an AUC value of 99.65%. This study presents a machine learning method that has been developed to address different technical challenges in the field of deepfake detection and effectively identify deepfakes. The proposed method has demonstrated successful performance compared to state-of-the-art models, proving its effectiveness in accurately detecting fake content within videos.

Deepfake is the general expression of fake images or videos. Images and videos created with Deep Neural Networks (DNN) can cause both personal and national problems. Deepfake technology has been making progress day by day. Accordingly, both the detection methods and the diversity of the data sets to be trained are increasing. In this article, datasets created in deepfake technology and applications that use datasets for deepfake detection are evaluated. The data sets used are produced with different models. There exist both traditional and deep learning-based detection methods for deepfake detection. Due to the recent malicious use of deepfake technology, detection methods and methods to be used in applications are of great importance. When the detection methods are compared, the accuracy of the deep learning-based detection method is higher for each data set. The method that provides the highest accuracy among two different data sets is the Xception method.

Software can be vulnerable to various types of interference. The production of cyber threat intelligence for closed source software requires significant effort, experience, and many manual steps. The objective of this study is to automate the process of producing cyber threat intelligence, focusing on closed source software vulnerabilities. To achieve our goal, we have developed a system called cti-for-css. Deep learning algorithms were used for detection. To simplify data representation and reduce pre-processing workload, the study proposes the function-as-sentence approach. The MLP, OneDNN, LSTM, and Bi-LSTM algorithms were trained using this approach with the SOSP and NDSS18 binary datasets, and their results were compared. The aforementioned datasets contain buffer error vulnerabilities (CWE-119) and resource management error vulnerabilities (CWE-399). Our results are as successful as the studies in the literature. The system achieved the best performance using Bi-LSTM, with F1 score of 82.4%. Additionally, AUC score of 93.0% was acquired, which is the best in the literature. The study concluded by producing cyber threat intelligence using closed source software. Shareable intelligence was produced in an average of 0.1 s, excluding the detection process. Each record, which was represented using our approach, was classified in under 0.32 s on average.

Cyber criminals can use many different methods to completely lose their traces after their actions or to complicate the work of forensic experts. Detection of such suspected anti-forensic activities can guide the detection of cyber crimes that have been experienced but not yet detected. One of the tools that can be used for this purpose is SIEM (Security Information and Event Management) systems. With proper configuration and management, these systems facilitate the work of security analysts in attack detection and incident response stages. In addition, it facilitates the discovery of evidence after a cybercrime and reduces evidence collection times. In this study, logs and log collection methods that can be produced by systems that are likely to be found in an enterprise infrastructure were examined. It has been researched how to detect sample anti-forensics attack types with the help of a selected SIEM application and the results are shown on a created use case template.