Penetration testing, a critical component of cybersecurity, typically requires extensive time and effort to find vulnerabilities. Beginners in this field often benefit from collaborative approaches with the community or experts. To address this, we develop Cybersecurity Intelligent Penetration-testing Helper for Ethical Researchers (CIPHER), a large language model specifically trained to assist in penetration testing tasks as a chatbot. Unlike software development, penetration testing involves domain-specific knowledge that is not widely documented or easily accessible, necessitating a specialized training approach for AI language models. CIPHER was trained using over 300 high-quality write-ups of vulnerable machines, hacking techniques, and documentation of open-source penetration testing tools augmented in an expert response structure. Additionally, we introduced the Findings, Action, Reasoning, and Results (FARR) Flow augmentation, a novel method to augment penetration testing write-ups to establish a fully automated pentesting simulation benchmark tailored for large language models. This approach fills a significant gap in traditional cybersecurity Q&A benchmarks and provides a realistic and rigorous standard for evaluating LLM’s technical knowledge, reasoning capabilities, and practical utility in dynamic penetration testing scenarios. In our assessments, CIPHER achieved the best overall performance in providing accurate suggestion responses compared to other open-source penetration testing models of similar size and even larger state-of-the-art models like Llama 3 70B and Qwen1.5 72B Chat, particularly on insane difficulty machine setups. This demonstrates that the current capabilities of general large language models (LLMs) are insufficient for effectively guiding users through the penetration testing process. We also discuss the potential for improvement through scaling and the development of better benchmarks using FARR Flow augmentation results.

Recent advancements in generative AI, particularly diffusion-based image editing, have enabled the transformation of images into highly realistic scenes using only text instructions. This technology offers significant potential for generating diverse synthetic datasets to evaluate model robustness in real-world applications, especially in the era of autonomous driving. In this paper, we introduce Cityscape-Adverse, a benchmark that employs diffusion-based image editing to simulate eight adverse conditions, including variations in weather, lighting, and seasons, while preserving the original semantic labels. We evaluate the reliability of diffusion-based models in generating realistic scene modifications and assess the performance of state-of-the-art CNN and Transformer-based semantic segmentation models under these challenging conditions. Additionally, we analyze which modifications have the greatest impact on model performance and explore how training on synthetic datasets can improve robustness in real-world adverse scenarios. Our results demonstrate that all tested models, particularly CNN-based architectures, experienced significant performance degradation under extreme conditions, while Transformer-based models exhibited greater resilience. We verify that models trained on Cityscape-Adverse show significantly enhanced resilience when applied to unseen domains. Code and datasets will be released at https://github.com/naufalso/cityscape-adverse.

Commodity processor architectures are releasing various instruction set extensions to support security solutions for the efficient mitigation of memory vulnerabilities. Among them, tagged memory extension (TME), such as ARM MTE and SPARC ADI, can prevent unauthorized memory access by utilizing tagged memory. However, our analysis found that TME has performance and security issues in practical use. To alleviate these, in this paper, we propose CoMeT, a new instruction set extension for tagged memory. The key idea behind CoMeT is not only to check whether the tag values in the address tag and memory tag are matched, but also to check the access permissions for each tag value. We implemented the prototype of CoMeT on the RISC-V platform. Our evaluation results confirm that CoMeT can be utilized to efficiently implement well-known security solutions, i.e., shadow stack and in-process isolation, without compromising security.

This research paper entails an examination of the Enhanced Wi-Fi protocol, focusing on its control command reverse-engineering analysis and subsequent demonstration of a hijacking attack. Our investigation discovered vulnerabilities in the Enhanced Wi-Fi control commands, rendering them susceptible to hijacking attacks. Notably, the study established that even readily available and cost-effective commercial off-the-shelf Wi-Fi routers could be leveraged as effective tools for executing such attacks. To illustrate this vulnerability, a proof-of-concept remote hijacking attack was carried out on a DJI Mini SE drone, whereby we intercepted the control commands to manipulate the drone's flight trajectory. The findings of this research emphasize the critical necessity of implementing robust security measures to safeguard unmanned aerial vehicles against potential hijacking threats. Considering that civilian drones are now used as war weapons, the study underscores the urgent need for further exploration and advancement in the domain of civilian drone security.

Recent advancements in generative AI, particularly diffusion-based image editing, have enabled the transformation of images into highly realistic scenes using only text instructions. This technology offers significant potential for generating diverse synthetic datasets to evaluate model robustness. In this paper, we introduce Cityscape-Adverse, a benchmark that employs diffusion-based image editing to simulate eight adverse conditions, including variations in weather, lighting, and seasons, while preserving the original semantic labels. We evaluate the reliability of diffusion-based models in generating realistic scene modifications and assess the performance of state-of-the-art CNN and Transformer-based semantic segmentation models under these challenging conditions. Additionally, we analyze which modifications have the greatest impact on model performance and explore how training on synthetic datasets can improve robustness in real-world adverse scenarios. Our results demonstrate that all tested models, particularly CNN-based architectures, experienced significant performance degradation under extreme conditions, while Transformer-based models exhibited greater resilience. We verify that models trained on Cityscape-Adverse show significantly enhanced resilience when applied to unseen domains. Code and datasets will be released at https://github.com/naufalso/cityscape-adverse.