AI Guardrails Draw Criticism from Offensive Security Researchers — techcrunch.com

Mobileye Set for Leadership Change as CEO Amnon Shashua Plans Exit Amid Robotaxi and Robotics Push — techcrunch.com

Alexa Plus update expands smart home reach with deeper device control — theverge.com

FCC Chairman Brendan Carr escalates pressure on broadcasters and ownership rules — theverge.com

LG Monitor App Pulls McAfee Pop-Up Ads After Microsoft Intervention — arstechnica.com

AI Guardrails Draw Criticism from Offensive Security Researchers — techcrunch.com

Mobileye Set for Leadership Change as CEO Amnon Shashua Plans Exit Amid Robotaxi and Robotics Push — techcrunch.com

Alexa Plus update expands smart home reach with deeper device control — theverge.com

FCC Chairman Brendan Carr escalates pressure on broadcasters and ownership rules — theverge.com

LG Monitor App Pulls McAfee Pop-Up Ads After Microsoft Intervention — arstechnica.com

AMD's Big AI Bet

AMD Launches Helios Rack-Scale AI System With MI455X GPU and 6th Gen EPYC Venice, Signs OpenAI and Anthropic as Customers

At Advancing AI 2026, AMD unveiled Helios — a rack-scale AI solution packing 72 MI455X GPUs (2nm, 432GB HBM4 each) and 18 EPYC Venice CPUs rated at 2.9 exaflops FP4 inference. OpenAI, Anthropic, Meta, Microsoft, and Oracle are confirmed customers. AMD separately announced a strategic partnership with Anthropic to deploy up to 2GW of MI450-series GPUs and a $5B equity investment.

AMD · 2026-07-23 · Read the dispatch →

AI chip startup Etched raises $300M Series C at $10.3B valuation

Sequoia-led round with a16z, Jane Street, SK hynix; $1B in booked inference chip orders positions Etched as a major ASIC challenger.

TechCrunch · 2026-07-23

xAI Launches Workflows in Grok Build — Parallel Agent Orchestration at Scale

Grok Build can now fan out tasks across up to 1,024 parallel agents, verify results, and report back in one background run using plain-language workflows saved as slash commands.

x.ai · 2026-07-23

SemiAnalysis: Vera Rubin NVL72 Delivers 5.4x More Tokens Per MW Than GB200

Running DeepSeek R1, NVIDIA's Vera Rubin achieves 5.4x more tokens per megawatt and 5x more per dollar vs GB200 NVL72, with a $3.57/GPU-hour TCO that makes it the only viable option for serving workloads at 350 tok/s/user.

SemiAnalysis · 2026-07-22

YouTube & Video

Google achieves quantum self-correction breakthrough — computers that retune themselves mid-computation

Google's quantum system learns from its own errors and retunes itself mid-computation, beating expert calibration and reaching record error rates without human intervention.

YouTube — Beyond AI News · 2026-07-23

Chinese labs face sanctions allegations over industrial-scale Kimi K-3 model distillation

Analysis covers investor fears about cheaper models undercutting OpenAI and Anthropic, circular financing risks, runaway CapEx, and inference bottlenecks.

YouTube — The AI Daily Brief · 2026-07-24

Apple preparing full Mac lineup overhaul to meet surging AI demand

Apple plans new versions of every Mac across fall 2026 and 2027, targeting demand for powerful AI-capable laptops and desktops across iMac, MacBook Pro, and MacBook Air lines.

YouTube — Bloomberg Technology · 2026-07-23

Hardware & Robotics

CoreWeave benchmarks NVIDIA Vera Rubin NVL72: 10x tokens per megawatt vs Grace Blackwell

First published benchmarks on DeepSeek R1 show 10x more throughput per megawatt compared to GB200 NVL72 at matched interactivity, a critical metric for power-constrained AI factories.

CoreWeave · 2026-07-22

AMD Launches Kria AI SOM and Robotics Developer Platform for Physical AI

Industry's first open, fully integrated robotics platform combining Ryzen AI Embedded X100 (16 Zen 5 cores, RDNA 3.5 GPU, dedicated NPU) on a compact COM-HPC module for autonomous machines.

UC Advanced · 2026-07-23

AMD Debuts Next-Generation AI Infrastructure for Frontier Models and Agentic Workloads

Full coverage of AMD's Advancing AI 2026 announcements including Helios, MI455X GPU, 6th Gen EPYC Venice on 2nm, and Kria robotics platform with OpenAI CEO Sam Altman joining Lisa Su on stage.

SiliconANGLE · 2026-07-23

Tools & Startups

Runway Launches Media Router — AI model selector by cost, speed, or quality

Runway's new tool automatically selects the best image, video, or audio generation model based on developer priorities, routing across the crowded generative media landscape.

TechCrunch · 2026-07-23

Echo delivers Fable-level quality at one-third the cost using open-weight models

An adaptive model built from a pool of open-weight models (GLM-5.2, Kimi K2.7, and others) that dynamically routes to the most cost-effective open model for each task.

Hacker News · 2026-07-23

PromptQL — multiplayer AI workspace from the Hasura team aims to replace Slack

An AI-native team workspace where shared threads replace channels, capturing tribal knowledge across databases, SaaS apps, and coding agents into a self-building business wiki.

Product Hunt · 2026-07-24

OpenCode Superapp — Codex with local self-hosted models and voice interaction

Unified coding assistant combining OpenAI Codex with local models for online and offline use with model flexibility.

Product Hunt · 2026-07-24

OneCLI — open-source credential gateway keeping secrets out of AI agents

Intercepts API keys before they reach AI agents, providing secure access to services without exposing credentials in context or logs.

Hacker News · 2026-07-23

In Brief

Source:

AMD Launches Helios Rack-Scale AI System With MI455X GPU and 6th Gen EPYC Venice, Signs OpenAI and Anthropic as Customers

AI chip startup Etched raises $300M Series C at $10.3B valuation

xAI Launches Workflows in Grok Build — Parallel Agent Orchestration at Scale

SemiAnalysis: Vera Rubin NVL72 Delivers 5.4x More Tokens Per MW Than GB200

YouTube & Video

Google achieves quantum self-correction breakthrough — computers that retune themselves mid-computation

Chinese labs face sanctions allegations over industrial-scale Kimi K-3 model distillation

Apple preparing full Mac lineup overhaul to meet surging AI demand

Hardware & Robotics

CoreWeave benchmarks NVIDIA Vera Rubin NVL72: 10x tokens per megawatt vs Grace Blackwell

AMD Launches Kria AI SOM and Robotics Developer Platform for Physical AI

AMD Debuts Next-Generation AI Infrastructure for Frontier Models and Agentic Workloads

Tools & Startups

Runway Launches Media Router — AI model selector by cost, speed, or quality

Echo delivers Fable-level quality at one-third the cost using open-weight models

PromptQL — multiplayer AI workspace from the Hasura team aims to replace Slack

OpenCode Superapp — Codex with local self-hosted models and voice interaction

OneCLI — open-source credential gateway keeping secrets out of AI agents

Money & Markets

Insurance startup Corgi reportedly raises B2 extension at $4B valuation

AegisAI lands $36M Series A to stop AI-driven spear phishing

Ropedia raises $30M pre-A for physical AI data infrastructure

Elio raises $21M for AI-native sensing and perception sensors

Prosper Medical raises $16M seed for AI-powered concierge primary care

In Brief

AMD Launches Helios Rack-Scale AI System With MI455X GPU and 6th Gen EPYC Venice, Signs OpenAI and Anthropic as Customers

AI chip startup Etched raises $300M Series C at $10.3B valuation

xAI Launches Workflows in Grok Build — Parallel Agent Orchestration at Scale

SemiAnalysis: Vera Rubin NVL72 Delivers 5.4x More Tokens Per MW Than GB200

YouTube & Video

Google achieves quantum self-correction breakthrough — computers that retune themselves mid-computation

Chinese labs face sanctions allegations over industrial-scale Kimi K-3 model distillation

Apple preparing full Mac lineup overhaul to meet surging AI demand

Hardware & Robotics

CoreWeave benchmarks NVIDIA Vera Rubin NVL72: 10x tokens per megawatt vs Grace Blackwell

AMD Launches Kria AI SOM and Robotics Developer Platform for Physical AI

AMD Debuts Next-Generation AI Infrastructure for Frontier Models and Agentic Workloads

Tools & Startups

Runway Launches Media Router — AI model selector by cost, speed, or quality

Echo delivers Fable-level quality at one-third the cost using open-weight models

PromptQL — multiplayer AI workspace from the Hasura team aims to replace Slack

OpenCode Superapp — Codex with local self-hosted models and voice interaction

OneCLI — open-source credential gateway keeping secrets out of AI agents

Money & Markets

Insurance startup Corgi reportedly raises B2 extension at $4B valuation

AegisAI lands $36M Series A to stop AI-driven spear phishing

Ropedia raises $30M pre-A for physical AI data infrastructure

Elio raises $21M for AI-native sensing and perception sensors

Prosper Medical raises $16M seed for AI-powered concierge primary care

Repos on GitHub

Trending on HF

In Brief