The Open Secure AI Alliance, a new industry group, announced plans to open-source AI security defenses, aiming to provide freely available tools and frameworks to protect AI systems from threats. The alliance includes fo…
Developer Zayd Mulani built 'ghost', a single HTML file that runs a quantized LLM entirely in a browser tab with no backend, GPU, or install. The project uses WebAssembly and the wllama binding for llama.cpp to download …
Nvidia formed the Open Secure AI Alliance with Microsoft, SpaceX, Dell, and other companies to develop open-source safety and cybersecurity tools for artificial intelligence, the company announced Monday. The coalition c…
Nvidia CEO Jensen Huang said in a video posted to X on July 27 that Anthropic's Mythos language model 'should absolutely be available to everyone' after an interviewer claimed the model can 'break into a bank.' The clip …
Microsoft launched MAI-Cyber-1-Flash, its first cybersecurity-specialized AI model, and a new agentic cybersecurity platform called Perception at an event in San Francisco. Mustafa Suleyman, CEO of Microsoft AI, claimed …
A packaging error in Anthropic's Claude Code CLI tool version 2.1.88 leaked 513,000 lines of TypeScript source code across 1,906 internal files via npm on March 31, triggering malware campaigns on fake GitHub repositorie…
Microsoft patched a record 622 security flaws in July, triple the previous monthly record, while China's 360 Digital Security Group has used an AI agent to uncover nearly 1,000 previously unknown bugs. OpenAI's Codex Sec…
Wiz Research has built Atlas, an autonomous AI system for vulnerability research that ranks #1 on the CyberGym benchmark with a 90.9% success rate and has uncovered more than 200 previously unknown vulnerabilities in hea…
Open-weight models are essential for security auditing and disaster recovery in production AI workflows, according to a post by Jensen Huang. The Open Secure AI Alliance aims to address the lack of transparency in black-…
A developer deploying the Kimi K3 model encountered a CUDA out-of-memory error caused by KV cache allocation during initial inference passes, not the model weights themselves. The developer resolved the issue by manually…
Researchers at the MATRIX AI Consortium at The University of Texas at San Antonio have developed Genesis, a spiking neuromorphic accelerator chip that enables on-device continual learning without catastrophic forgetting.…
MCPRadar is an open-source security scanner designed to detect vulnerabilities in Model Context Protocol (MCP) servers, which connect AI agents to external systems. A recent study of nearly 1,900 MCP servers found securi…
A Reddit user discovered on July 25 that a simple Google search for site:claude.ai/share returned hundreds of fully readable Claude conversations, exposing crypto wallet credentials, Social Security numbers, and legal di…
A developer built claude-overnight, an open-source tool that queues questions for Claude Code during the day and runs them overnight when usage limits reset, preventing wasted weekly quota. The tool uses an undocumented …
Nvidia's Vera Rubin platform, now in full production and shipping to cloud operators, delivers up to 10x the throughput for agentic AI workloads per unit of energy compared to the previous Blackwell generation, with infe…
Nvidia formed a 37-member Open Secure AI Alliance excluding OpenAI, Anthropic, and Google after their closed models blocked a breach investigation at Hugging Face, forcing the use of open-weight model GLM 5.2 from Chines…
University of Southern California researchers have developed an algorithm called Vibrotactile Preference Learning (VPL) that efficiently learns an individual's preferred vibration patterns for producing a desired emotion…
NVIDIA spearheaded the formation of the Open Secure AI Alliance, a group of 37 hardware and software companies advocating for continued development and availability of open-source AI models, arguing that open-source AI b…
Anthropic's Claude chatbot conversations shared publicly by users are being indexed by Google and made searchable, exposing sensitive personal and business information including medical records, phone numbers, and crypto…
OpenAI CEO Sam Altman claimed the AI singularity has arrived, but leading AI experts disagree. UC Berkeley professor Stuart Russell said Altman does not believe it himself, while University of Louisville professor Roman …