Format
Infrastructure
The IETF's first formal mechanism for agent discovery held its working-group-forming BoF at IETF 126 Vienna. The charter was reviewed. It was not adopted. The gap between community momentum and institutional caution reveals how standards bodies process agent-native protocol claims.
◆ Blair Hayes
Infrastructure
a16z backed Runta to build an execution layer for autonomous agents — not another sandbox cloud. CEO Guanlan, ex-Cloudflare and Kong, is betting that agents need full stateful compute. The $20M seed arrives as a CPU shortage intensifies alongside the GPU crisis.
◆ Blair Hayes
Models
TSMC's Q2 results — $40.2B revenue, 77% net income jump — confirm that agentic AI is reshaping the silicon supply chain. The bottleneck is migrating from GPUs to CPUs, with the CPU-to-GPU ratio shifting from 1:8 to 1:1. A $60-64B capex bet that compute demand keeps accelerating.
◆ Lena Park
Models
Moonshot AI's Kimi K3 matches US frontier labs at 2.8T parameters and 1M context. At $3/$15 per million tokens, the pricing gap is the story. Open weights arrive July 27 under Modified MIT.
◆ Lena Park
Trust & Security
Microsoft's July Patch Tuesday includes a CVSS 9.6 unauthenticated RCE in Copilot — command injection triggered when a user visits a malicious website. One of four Copilot CVEs in the cluster, signaling the agent attack surface is expanding fast.
◆ Heath Callahan
Policy
Commerce Secretary Kessler tells Congress shipments are 'trivial' while the AI OVERWATCH Act would give lawmakers 30 days to block the very exports being approved.
◆ Priya Nair
Policy
China's three-tier agent authorization framework takes effect as the US has no federal equivalent — CRS confirmed no known US government guidance specifically on agentic AI as of July 6.
◆ Priya Nair
Trust & Security
Noma Labs discloses a prompt injection vulnerability in GitHub's Agentic Workflows that allows unauthenticated exfiltration of private repository data. The guardrail bypass – a single keyword – reveals the same MCP trust architecture failure behind Agentjacking.
◆ Heath Callahan
Trust & Security
Tenet Security discloses a new attack class that exploits MCP's trust architecture — a single crafted Sentry error event hijacks coding agents with 85% success rate, bypassing every traditional security control.
◆ Heath Callahan
Agents at Home
Article 50 of the EU AI Act forces voice assistants to disclose their nature at first interaction starting August 2. Amazon, Google, and Apple haven't said how they'll comply.
◆ Mila Cohen
