NVIDIA Open Secure AI Alliance Launches — 37 Companies Open-Source NOOA Agent Security Stack
NVIDIA's 37-member Open Secure AI Alliance open-sources NOOA for auditable AI agents — conspicuously missing OpenAI, Google, and Anthropic.
News
Anthropic and Cognizant Expand to Global Premier Partnership — 350K Associates, Claude Code in Enterprise
Cognizant becomes Anthropic Global Premier Partner, embedding Claude Code and MCP into Flowsource for 350K associates.
Anthropic's Official Position on Open-Weights AI Models
Dario Amodei clarifies Anthropic never backed open-weight bans — targets chip export controls and distillation instead.
DeepSeek Pauses $71B Fundraising Round After Liang Wenfeng Compute-Gap Remarks Leak
DeepSeek pauses $71B funding round after Liang Wenfeng's candid compute-gap remarks leak — what it means for open-weight AI deployments.
Inside the Token Relay Market: How Gray-Market LLM API Resellers Operate — And Why It Matters for Agentic Teams
Gray-market LLM API relays offer 90%+ discounts but audits find ~46% serve wrong models and some inject malicious payloads — a risk map for agentic teams.
Cekura Voice Benchmark: ElevenLabs Wins Latency at 1.73s, Retell Wins Reliability at 96.6%
Cekura benchmarks 6 voice AI platforms: ElevenLabs wins p50 latency (1.73s), Retell wins reliability (96.6% pass rate) over 1,000+ test turns.
CrewAI 1.15.6 Released: Anthropic Tool-Use Block Detection Fixed for Claude Opus 5 and Fable 5
CrewAI 1.15.6 fixes Anthropic preview tool-use block detection for Claude Opus 5/Fable 5 users plus three other targeted stability fixes.
Ant International ANTOM Copilot Upgrades: Agent Trust Rating, KYA Framework, WorldFirst Enterprise Accounts
Ant International's ANTOM upgrades its agent commerce stack with trust ratings and KYA compliance — Asia's answer to x402 and Stripe's agent payment push.
Cloudflare Content Independence Day 2: Granular AI Traffic Controls — Search, Agent, and Training Bots
Cloudflare formalizes agent bots as a distinct traffic class. Here's what it means for teams running web-browsing AI agents.
Inflect-Micro-v2: Complete Text-to-Speech in 9.36M Parameters — Edge TTS for AI Agents
A complete English TTS model in 37MB: Inflect-Micro-v2 brings real-time speech synthesis to edge hardware and lightweight agent interfaces.
Open-Weight AI Is Having Its Kubernetes Moment — And the US Shouldn't Wall Itself Off
D2iQ co-founder Tobi Knaup argues open-weight AI models are following Kubernetes' path as neutral infrastructure — and US restrictions would repeat Kubernetes-era mistakes.
HubSpot Launches Agent Hub and Agent Builder in Public Beta — Unified Console for Enterprise AI Agent Management
HubSpot debuts Agent Hub and Agent Builder in public beta — a unified no-code console for managing AI agents across its 150K+ enterprise CRM customers.
IETF Convenes 'agentproto' BoF in Vienna — Standards Body Takes on the MCP vs. A2A vs. ACP Protocol Fragmentation Fight
IETF formally convenes to address AI agent protocol fragmentation — MCP, A2A, ACP, and more face standardization scrutiny in Vienna.
Anthropic Releases Claude Opus 5 — New SOTA for Agentic Coding, Knowledge Work, and Computer Use
Claude Opus 5 launches July 24 as Anthropic's new agentic SOTA — #1 on AA Intelligence Index, double the coding score of Opus 4.8, at half Fable 5's cost.
Cognition Acquires The Interaction Company — Devin Maker Buys Consumer iMessage AI Agent 'Poke'
Cognition acquires Poke — Devin's maker buys the only Apple-approved iMessage AI agent, signaling a push from enterprise coding into always-on personal agents.
Kimi K3 AI Agents Find 19 Redis Zero-Days in 90 Minutes — Multi-Agent Security Exploit Demonstration
32 Kimi K3 sub-agents found 19 Redis zero-days in 90 minutes — a landmark (if caveated) demo of multi-agent offensive security at scale.
Nvidia, Microsoft, Meta Lead 25-Org Coalition Against Open-Weight AI Restrictions — 'Open Weights and American AI Leadership'
Nvidia, Microsoft, Meta and 22 others sign 'Open Weights and American AI Leadership' — Jensen Huang's first X post amplifies the call against AI model restrictions.
200 Startup Founders Urge U.S. Government Not to Ban Chinese Open-Weight AI Models
200 startup founders push back on a potential U.S. ban of Chinese open-weight AI models like Kimi K3 and GLM-5.2, warning it would kill local AI innovation.
DARPA and U.S. Air Force Fly AI-Controlled F-16 in First Autonomous Agent Combat Test
DARPA's VENOM program flew an AI-controlled F-16 in the first autonomous agent combat flight test — AI agents are moving into the physical world.
Google Releases AI & Economy ATLAS v1.0 — 15 Million Real Gemini Interactions Mapped Across 150 Countries
Google's AI & Economy ATLAS v1.0 maps 15M real Gemini interactions across 150 countries — AI overwhelmingly aids collaboration, not automation.
Screenpipe (YC S26) Launches — Record Your Work, Turn It Into AI Agents
Screenpipe (YC S26) records your screen and audio locally, giving AI agents like OpenClaw real work context via MCP — no cloud required.
Trump Administration Threatens Sanctions on Chinese AI Models Over IP Theft — 200 Startups Push Back
Trump administration weighs sanctions on Chinese AI labs for IP theft. 200 startups via Little Tech Association urge restraint, warning against broad model restrictions.
Kimi K3 Open Weights Drop July 27 — Prepare to Run the 2.8T MoE Locally
Kimi K3's 2.8T MoE open weights drop July 27—here's how to prepare your hardware and vLLM setup for local deployment of the agentic giant.
Microsoft Agent Framework Harness Released — Batteries-Included Runtime for Autonomous .NET and Python Agents
Microsoft's Agent Framework Harness goes stable: batteries-included agent loop, context compaction, memory, and approvals for .NET and Python agents.
Orbit v0 Released — Open-Source Multi-Agent Safety Evaluation Framework from UK AISI / Wittlab
Orbit v0 launches: open-source multi-agent safety framework with 11 evaluation scenarios, threat models, and defenses built on UK AISI's Inspect AI.