Your daily source for AI news, tools reviews, and practical guides to artificial intelligence
A fraudulent Hugging Face repository impersonating an OpenAI release delivered infostealer malware to Windows systems, accumulating approximately 244,000 downloads before removal. The attack highlights growing security risks within public AI model registries.
A deceptive repository on Hugging Face imitating an OpenAI release infected tens of thousands of Windows machines with information-stealing malware, raising concerns about security risks in AI model registries.
The European Union’s efforts to regulate artificial intelligence face obstacles as key AI developers OpenAI and Anthropic differ in their willingness to grant regulators access to their latest models, revealing the limits of EU oversight.
Mozilla has harnessed Anthropic’s Claude Mythos Preview, an agentic AI system, to detect 271 previously unknown vulnerabilities in Firefox 150, including bugs persisting for up to two decades. This AI-driven approach automates testing and promises enhanced security in future code commits.
As Clawdbot becomes a dominant AI chatbot platform in business operations, understanding its security vulnerabilities is crucial. This article explores the primary risks associated with Clawdbot and practical measures to mitigate them in 2026.
APIs and MCPs are essential technologies enabling data exchange in software and AI environments. While APIs facilitate communication between software applications, MCPs are tailored for large language models to access tools and data efficiently. This article explores their distinctions, use cases, and the importance of gateways for security in AI-driven systems.
At the Google Cloud Next conference, NVIDIA and Google introduced the A5X bare-metal instances powered by NVIDIA Vera Rubin NVL72 systems, promising up to tenfold reductions in AI inference costs and significant performance gains. This new architecture addresses the scalability and security challenges of deploying AI workloads at an unprecedented scale.
At the Google Cloud Next event, Google and NVIDIA unveiled a new hardware and software architecture aimed at drastically reducing the cost and increasing the efficiency of AI inference at scale, promising up to tenfold improvements in cost per token and token throughput.
Nicholas Moore infiltrated multiple U.S. government networks using stolen credentials and publicly exposed sensitive personal data on social media, leading to his recent sentencing to probation.
Although recently flagged by the Pentagon as a supply-chain risk, AI company Anthropic continues to maintain communications with senior figures from the Trump administration, indicating a potential warming in their relationship.