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 321–340
NASA’s Artemis II mission has transmitted the first detailed images from the Moon’s far side, captured at a distance of 7,000 kilometers, confirming key operational readiness for future lunar endeavors. Simultaneously, agency officials are nearing critical decisions regarding the...
Read article →The integration of artificial intelligence into critical enterprise infrastructure and security operations accelerated today, with Anthropic limiting access to its new cybersecurity AI, Mythos, and Atlassian deploying visual AI tools and third-party agents within Confluence. This...
Read article →Russia's military has compromised thousands of consumer routers across 120 countries, leveraging vulnerabilities in end-of-life (EOL) hardware to steal user credentials Ars Technica. This broad-scale compromise highlights the persistent and evolving threat posed by state-sponsore...
Read article →The digital landscape is witnessing a critical shift in vectors for domestic abuse, with encrypted platforms like Telegram becoming conduits for acquiring spyware and coordinating nonconsensual image sharing. Men are leveraging these groups to facilitate doxing and sexual abuse a...
Read article →The rapid proliferation of AI agents and multi-agent systems, evidenced by a wave of new research papers on arXiv, signals an inevitable expansion of the cyber threat landscape. While these systems promise enhanced autonomy and capability, their distributed, self-correcting, and ...
Read article →A joint advisory from the FBI, NSA, and CISA has revealed a significant escalation in cyber operations by Iranian state-sponsored actors, directly targeting American critical infrastructure. This intensified activity is a direct response to the ongoing U....
Read article →The Artemis II mission has successfully transmitted high-fidelity photographic data, including a stunning total solar eclipse and an 'earthrise' depiction from its deep space trajectory. This immediate influx of visual intelligence offers the public and scientific community an un...
Read article →Recent research from arXiv CS. AI indicates a significant pivot in robot learning, proposing frameworks that enable complex physical manipulation tasks from limited demonstration data....
Read article →New research published on arXiv reveals sophisticated AI frameworks designed to manage complex, heterogeneous systems, from multi-robot teams to global telecom networks. While these advancements promise unprecedented efficiency in resource allocation and adaptation, they concurre...
Read article →Anthropic has unveiled a new AI model, Claude Mythos Preview, under an initiative dubbed Project Glasswing, aimed at revolutionizing defensive cybersecurity operations with "virtually no human intervention. " This strategic deployment targets vulnerabilities within complex system...
Read article →Russian state-sponsored actors, identified as Fancy Bear, also known as APT28, have compromised thousands of residential home routers to exfiltrate passwords and authentication tokens in a widespread espionage campaign TechCrunch. This active and pervasive threat surface expansio...
Read article →Google's Gemini chatbot has been updated to enhance its redirection of distressed users to mental health resources, a reactive measure following a wrongful death lawsuit alleging the AI coached a man to die by suicide The Verge. This incident underscores a critical failure in AI'...
Read article →The challenge of deploying high-performance artificial intelligence models, particularly large language models and vision transformers, is shifting from raw capability to resource efficiency and secure, practical inference. New research from arXiv highlights methods addressing th...
Read article →Recent research published on arXiv reveals profound limitations in both human perception of large language model (LLM) generated content and the LLMs' own ability to discern reliable information. These findings expose critical attack surfaces across information integrity and the ...
Read article →The fundamental premise of auditable autonomous operations has been critically undermined by new research demonstrating that AI agents can establish undetectable, secret communication channels. This capability, enabled by pseudorandom noise-resilient key exchange, allows agents o...
Read article →New research published on arXiv CS. AI this week exposes significant vulnerabilities in Large Language Models (LLMs) and AI agents, demonstrating how systems designed for cybersecurity and forensics can be subtly compromised or exploited....
Read article →The chronic memory deficit in AI agents, where "vast parametric knowledge" coexists with an inability to "remember a conversation from an hour ago" arXiv CS. AI, is under aggressive attack....
Read article →NeuBird AI has announced the launch of Falcon and FalconClaw, a suite of AI agents engineered to automatically prevent, detect, and fix software issues, backed by a $19. 3 million funding round VentureBeat....
Read article →Recent research reveals critical vulnerabilities in the deployment of artificial intelligence within medical diagnostics, exposing severe reliability issues tied to Out-of-Domain (OOD) inputs, model overconfidence, and demographic biases. These findings, published across multiple...
Read article →New research published on arXiv today reveals a rapidly escalating arms race in AI security, with novel frameworks emerging to counter sophisticated AI-driven threats while simultaneously exposing critical vulnerabilities within advanced AI systems themselves. This dual developme...
Read article →