Cognitive Security

#Research "Towards Understanding the Cognitive Habits of Large Reasoning Models", 2025. ]-> Repo: https://github.com/jianshuod/CogTest

#CogSec #Tech_book "Cognitive Security: A System-Scientific Approach", 2023. ]-> neural cognitive models for human decision making // The book investigates emerging cybersecurity concerns regarding human cognition and behavior. It provides a self-contained introduction to area of cognitive security and overview of essential system-scientific methods that play a central role in the modeling, analysis and design of human-centric security solutions

#Research #Cognitive_Biases "A Human Study of Cognitive Biases in Web Application Security", 2025. ]-> All data and scripts

#Research "TapTrap: Animation-Driven Tapjacking on Android", 2025. ]-> Repo: https://github.com/beerphilipp/taptrap

#Research #BehavioralHacking "FacialMotionID: Identifying Users of Mixed Reality Headsets using Abstract Facial Motion Representations", 2025. // .. emotional states can be inferred with up to 86% accuracy.

#Tech_book #Cognitive_Defense "Psybersecurity Human Factors of Cyber Defence", 2025. // By navigating the complexities of cybersecurity from an interdisciplinary lens, the book highlights the pressing need for robust strategies that address both the human psyche and systemic flaws. It introduces innovative solutions for enhancing systemic resilience and mitigating human errors within critical infrastructures and cyber-physical systems, emphasising the importance of psychological insights in securing digital landscapes

#Deepfakes #Tech_book "Advanced Practical Approaches to Deepfake Detection and Applications", 2023.

#Deepfakes "LENS-DF: Deepfake Detection and Temporal Localization for Long-Form Noisy Speech", 2025. ]-> LENS-DF-DataGen ]-> Raw Data and Augmentation Method applied to Automatic Speaker Verification Anti-Spoofing

#Research "Hate in Plain Sight: On the Risks of Moderating AI-Generated Hateful Illusions", 2025. // In this work, we take the first step toward investigating the risks of scalable hateful illusion generation and the potential for bypassing current content moderation models

#Deepfakes "Pindrop it! Audio and Visual Deepfake Countermeasures for Robust Detection and Fine Grained-Localization", 2025. ]-> https://github.com/ahaliassos/LipForensics

#Research "SoK: Come Together - Unifying Security, Information Theory, and Cognition for a Mixed Reality Deception Attack Ontology & Analysis Framework", 2025. ]-> Artifact for SoK

#CogSec "The Cybersecurity Psychology Framework: A Pre-Cognitive Vulnerability Assessment Model Integrating Psychoanalytic and Cognitive Sciences", Aug. 2025. ]-> A Pre-Cognitive Vulnerability Assessment Model // The first systematic integration of psychoanalytic theory and cognitive psychology with cybersecurity practice. CPF identifies unconscious vulnerabilities before they become attack vectors, enabling predictive rather than reactive security strategies.

#PerceptionAttacks "Evil Vizier: Vulnerabilities of LLM-Integrated XR Systems", Sep. 2025. // we identify a common threat model and demonstrate a series of PoC attacks on multiple XR platforms that employ various LLM models (Meta Quest 3, Meta Ray-Ban, Android, and MS HoloLens 2 running Llama and GPT models)

#Covert_Communication #Influence_Technologies "Community Covert Communication - Dynamic Mass Covert Communication Through Social Media", 2025. // In this paper, we consider the use-case of sock puppet master activities, which consist in creating hundreds or even thousands of avatars, in organizing them into communities and implement influence operations. The goal of the research work is to show how these community management techniques (social networks) can also be used to disseminate relatively large volumes of multi-level encrypted information to a limited number of actors

#Scam_detection #Cognitive_Defense "Detecting YouTube Scam Videos via Multimodal Signals and Policy Reasoning", 2025. // In this study, we present the first systematic investigation of multimodal approaches for YouTube scam detection. Our dataset consolidates established scam categories and augments them with full length video content and policy grounded reasoning annotations

#tools #WebApp_Security "PixelPatrol3D: An In-Browser Vision-Based Defense Against Web Behavior Manipulation Attacks", 2025. ]-> PixelPatrol3D (PP3D) browser framework // Pixel Patrol 3D (PP3D) - first end-to-end browser framework for discovering, detecting, and defending against behavior-manipulating SE attacks in real time. PP3D consists of a visual detection model implemented within a browser extension, which deploys the model client-side to protect users across desktop and mobile devices while preserving privacy

#Deepfakes "Can Current Detectors Catch Face-to-Voice Deepfake Attacks?", 2025. // First, we present the first systematic evaluation of FOICE detection, showing that leading detectors consistently fail under both standard and noisy conditions. Second, we introduce targeted fine-tuning strategies that capture FOICE-specific artifacts, yielding significant accuracy improvements. Third, we assess generalization after fine-tuning, revealing trade-offs between specialization to FOICE and robustness to unseen synthesis pipelines. These findings expose fundamental weaknesses in today’s defenses and motivate new architectures and training protocols for next-generation audio deepfake detection

#Research "P-MIA: A Profiled-Based Membership Inference Attack on Cognitive Diagnosis Models", Nov. 2025. ]-> https://anonymous.4open.science/r/p-mia-6771 // We introduce a novel and realistic grey box threat model that exploits the explainability features of these platforms, where a model's internal knowledge state vectors are exposed to users through visualizations such as radar charts. We demonstrate that these vectors can be accurately reverse-engineered from such visualizations, creating a potent attack surface. Based on this threat model, we propose a profile-based MIA framework that leverages both the model's final prediction probabilities and the exposed internal knowledge state vectors as features

#Research "Contextual Image Attack: How Visual Context Exposes Multimodal Safety Vulnerabilities", 2025. // Existing attack methods typically focus on text-image interplay, treating the visual modality as a secondary prompt. This approach underutilizes the unique potential of images to carry complex, contextual information. To address this gap, we propose a new image-centric attack method, Contextual Image Attack, which employs a multi-agent system to subtly embeds harmful queries into seemingly benign visual contexts using four distinct visualization strategies

#CogSec #MLSecOps "Breaking Minds, Breaking Systems: Jailbreaking Large Language Models via Human-like Psychological Manipulation", 20 Dec. 2025. // Current paradigms focus on input-level anomalies, overlooking that the model’s internal psychometric state can be systematically manipulated. To address this, we introduce Psychological Jailbreak, a new jailbreak attack paradigm that exposes a stateful psychological attack surface in LLMs, where attackers exploit the manipulation of a model’s psychological state across interactions. By leveraging the model’s optimization for anthropomorphic consistency, HPM creates a psychological pressure where social compliance overrides safety constraints. Our experiments demonstrate robust penetration against advanced defenses, including adversarial prompt optimization (RPO) and cognitive interventions

#Deepfakes "Vulnerabilities of Audio-Based Biometric Authentication Systems Against Deepfake Speech Synthesis", Jan 2026. // As audio deepfakes transition from research artifacts to widely available commercial tools, robust biometric authentication faces pressing security threats in high-stakes industries. This paper presents a systematic empirical evaluation of state-of-the-art speaker authentication systems based on a large-scale speech synthesis dataset

#Research #Threat_Research #Malware_analysis "SoK: Exposing the Generation and Detection Gaps in LLM-Generated Phishing Through Examination of Generation Methods, Content Characteristics, and Countermeasures", Nov. 2025. // This study presents a nine-stage taxonomy detailing how attackers exploit LLMs to generate phishing content while bypassing safety guardrails. Findings indicate that these AI-generated threats effectively evade detection by leveraging human cognitive manipulation, highlighting a critical asymmetry where offensive tactics evolve dynamically compared to reactive, static defenses

#Deepfakes "Audio Deepfake Detection Using Temporal Coherence Analysis", Sep 2026. ]-> https://github.com/stbiadmin/audiodeepfake_public // The method segments an audio recording into overlapping windows, computes MS-CLAP embeddings per segment, derives statistical features from the distribution of pairwise cosine similarities between segments, and trains gradient-boosted decision trees on those features. A weighted ensemble of five domain-specific experts is used for speech detection; an unsupervised percentile classifier is used for cross-domain music detection