Cybersecurity & Exploits

Global: Study Finds Over One Quarter of AI Agent Skills Contain VulnerabilitiesA new arXiv paper reports that 26.1% of AI agent skills collected from two leading marketplaces exhibit at least…

Global: New AI Model Achieves 99.44% Accuracy in Android Malware ClassificationA team of researchers has introduced a novel machine‑learning approach that combines the Fast Gradient Sign Method with a Diluted…

Global: Study Finds Black-Box Method Bypasses Political Safety Filters in Text-to-Image ModelsResearchers at a leading academic institution have demonstrated that a newly devised black-box framework can evade political safety filters…

Global: Inference Attacks on Web and Research Agents Using Network MetadataResearchers have demonstrated that language‑model‑based Web and Research Agents (WRAs), when deployed locally for privacy‑sensitive tasks, can be compromised by…

Global: Single-Shot Planning Offers New Defense for Computer Use Agents Against Prompt InjectionA team of researchers detailed a novel architectural approach that aims to shield AI-driven computer use agents from…

Global: New White-Box Adversarial Attack Leverages SHAP Values to Challenge Vision ModelsA paper submitted to arXiv on January 15, 2026 introduces a white‑box adversarial evasion technique that exploits SHAP (SHapley…

Global: Researchers Propose ‘Promptware’ Kill Chain for LLM AttacksA study released on arXiv in January 2026 outlines a new classification for threats against large language model (LLM) applications, coining the…

Global: AI-Driven Framework WAMM Achieves Near‑Perfect Accuracy in Web Attack DetectionA novel AI‑driven framework called WAMM has demonstrated the ability to classify web‑based attack traffic with up to 99.59% accuracy,…

Global: MindGuard Introduces Decision-Level Guardrails to Counter Tool Poisoning in LLM AgentsBackgroundA recent study published on arXiv outlines a new defensive framework aimed at mitigating tool poisoning attacks (TPA) that…

Global: Unsupervised Multi-View Anomaly Detection Achieves Over 99% Accuracy on IEC 61850 GOOSE TrafficOverviewResearchers have introduced an unsupervised, explainable anomaly detection system for IEC 61850 Generic Object‑Oriented Substation Event (GOOSE)…