Security Correspondent
Nadia Vale covers security failures, system safeguards and the difficult business of establishing trust. Her beat connects technical disclosures with their consequences for users and institutions. She favors specific threat models over sweeping claims that a system is safe or unsafe.
1,019 published articles · Showing 81–100
The rapid evolution of AI in computer vision, particularly in video and 3D generation, is marked by today's simultaneous release of seven research papers on arXiv. These documents detail advancements in long-form video synthesis, human animation, 3D scene editing, and pose estima...
Read article →Recent research from arXiv CS. AI unveils a rapid advancement in autonomous AI agent capabilities, particularly in complex reasoning, operational planning, and environmental interaction....
Read article →A new machine learning framework, Temporal Representation and Classification of Exploits (TRACE), has been proposed to enhance proactive Cyber Threat Intelligence (CTI) by identifying organizations likely to be targeted by cyberattacks. This development, detailed in a recent arXi...
Read article →Two distinct AI frameworks, BiTrajDiff and ContextFlow, have surfaced on arXiv, signaling a critical advancement in the precise generation and inference of complex trajectories. While these models promise enhanced predictive capabilities across domains from reinforcement learning...
Read article →New research from arXiv extends artificial intelligence's predictive capabilities to multimodal data, integrating complex information streams like images and unstructured text. While promising richer insights for applications from smart cities to automated systems, these advancem...
Read article →Recent research published on arXiv details significant advancements in AI for speaker recognition and verification, promising a new era of audio-first agents but simultaneously broadening the digital identity attack surface. One system achieved a Minimum Detection Cost Function (...
Read article →A new algorithm, ToMAToMP, for topological clustering has been introduced on arXiv, asserting "robustness guarantee[s]" and efficiency in data analysis arXiv CS. LG....
Read article →X, the platform formerly known as Twitter, is currently navigating a complex, bifurcated legal and regulatory landscape that underscores the inherent contradictions in modern digital governance. Simultaneously, the company fights to protect the anonymity of users targeted by a de...
Read article →New research published on arXiv reveals critical vulnerabilities in advanced AI applications for data management, from inventory control to the nascent field of AI agent social networks. The concurrent publications highlight fundamental challenges in data integrity, model robustn...
Read article →Three foundational research preprints, all published or updated on May 15, 2026, on arXiv CS. LG, signal a significant theoretical shift in machine learning....
Read article →Newly published research reveals critical vulnerabilities across advanced AI systems, from autonomous vehicle perception to the core robustness of foundation models and the inherent fairness of algorithmic decision-making. These findings, detailed in a series of pre-print papers ...
Read article →The release of TabPFN-3 marks a critical advancement in foundation models for tabular data, enabling state-of-the-art performance on datasets up to 1M training rows with reduced processing times arXiv CS. LG....
Read article →Recent academic publications on arXiv CS. LG, released on 2026-05-15, detail a wave of advancements in Large Language Model (LLM) efficiency, reasoning, and foundational architecture, alongside initial attempts at privacy-preserving unlearning....
Read article →Raindrop AI has today launched "Workshop," an open-source, MIT Licensed tool designed for local debugging and evaluation of AI agents VentureBeat. This development provides developers with direct, local visibility into agent operations, a critical step towards understanding and p...
Read article →The digital battlefield continues its relentless expansion, with two distinct but convergent threats emerging today. Russian government-backed operatives have been exposed attempting to compromise the Signal accounts of a prominent spyware investigator, a sophisticated maneuver r...
Read article →AMD is deploying its FSR 4. 1 upscaling technology to RDNA 3 and RDNA 2 graphics processing units (GPUs), a critical move to extend the operational lifespan and performance viability of existing hardware against the escalating demands of modern software The Verge....
Read article →While financial milestones like Clio's $500 million annual recurring revenue (ARR) and Mind Robotics' latest $400 million funding round are often celebrated as indicators of market validation, they concurrently represent a significant expansion of the global digital attack surfac...
Read article →Two new pre-print papers released on arXiv CS. LG outline critical advancements in efficient AI inference, tackling significant operational and privacy challenges inherent in contemporary machine learning deployments....
Read article →Recent research published on arXiv CS. AI details significant advancements in quantum computing, specifically concerning controllable quantum memory and the efficiency of quantum combinatorial optimization algorithms....
Read article →The efficacy of machine learning-based classifiers designed to detect deepfake audio content is now under critical scrutiny, following the introduction of DeePen, a novel penetration testing methodology. This development, detailed in a recent arXiv publication, directly challenge...
Read article →