Nvidia launches the Open Agent Safety Platform to quarantine AI agents that breach their boundaries within milliseconds, using OpenShell on its Vera AI CPU and a separate Sentry chip to monitor and enforce access limits.
China is considering approving Nvidia RTX Pro 5500 chip imports for Alibaba and ByteDance, with ByteDance prepared to order 1 million units—potentially reopening millions of sales previously restricted by Beijing.
NVIDIA launches the Open Agent Safety Platform, pairing OpenShell with the Sentry reference design on BlueField-4 DPUs to independently monitor and quarantine AI agents within milliseconds, while Salesforce brings approval and audit controls into Slack.
NVIDIA launches the Open Agent Safety Platform, pairing open-source OpenShell sandboxes with NVIDIA Sentry and BlueField-4 DPUs to enforce verifiable AI-agent rules on files, networks, tools, processes and credentials before and during execution.
Qualcomm says its Snapdragon 8 Elite Extreme Gen 6 can run 30-billion-parameter Mixture-of-Experts AI models directly on phones by loading needed experts from flash storage instead of requiring roughly 32GB of RAM.
Qualcomm launches the Snapdragon 8 Elite Extreme Gen 6, a mobile chip designed to run 30-billion-parameter mixture-of-experts AI models directly on phones, targeting next year’s Android flagships alongside Snapdragon Sound Elite Gen 2 wearables hardware.
Inferact open-sources TPU v7 megakernels and claims its Kimi K3 implementation surpasses 700 tokens per second with speculative decoding on 16 TPU v7 chips, using one Pallas kernel to fuse all 92 MoE layers.
Meta unveils the Charm AI pendant and expands its Muse assistant with voice, interactive avatars, Mac control, email and AI-glasses access, while Mark Zuckerberg says the 2-inch-display wearable packs 5G, dual cameras and a fingerprint sensor.
Perplexity brings its Portable on-device AI agents to AMD Ryzen AI Max Windows PCs, letting Pro and Max subscribers run PPLX 27B or Qwen 27B locally with Gmail, Outlook, Slack, and GitHub integrations without using Computer credits.
AMD says its sixth-generation EPYC processors lead agentic AI data-center workloads, offering up to 512 threads, 16 memory channels, PCIe Gen 6 and boost clocks up to 5 GHz for AI agent host nodes and sandboxes.