DAILY INTELLIGENCE BRIEF

The Vault - AI Edition

Wednesday, July 29, 2026
15 stories
12,885 words
58 min read

Business

5
05

Cyera Buys Oasis Security for About $1 Billion as AI-Agent Security M&A Heats Up

Cyera, the data-security company that only weeks ago raised $600 million at a $12 billion valuation, is buying its way into AI-agent security.

4 minRead →
06

U.S. to Award GlobalFoundries $300 Million to Speed Up AI Chip-to-Chip Connections

For years, the semiconductor arms race has been measured in nanometers. The next bottleneck is moving data between chips.

4 minRead →
07

Nvidia-Backed ChipAgents Raises $60 Million to Put AI Agents in Charge of Chip Design

The chips that power the AI boom are notoriously slow to design. ChipAgents wants autonomous agents to do the heavy lifting.

4 minRead →
08

Google and KDDI Launch an AI Startup Program to Grow Japan's Homegrown Founders

Japan punches above its weight in robotics and precision manufacturing, yet has struggled to produce dominant genAI startups. Google and KDDI want to change that.

4 minRead →
09

SoftBank Explores a $500 Million Deal for Swiss Construction-Robot Startup Gravis

SoftBank spent eight years building then dismantling its stake in Boston Dynamics. Now it is eyeing a different kind of robot.

3 minRead →

Llms Genai

3
10

Google's Gemini Omni Flash Brings Conversational Video Editing to Its Enterprise Agent Platform

Google has quietly reframed one of the messiest jobs in digital production — video editing — as a conversation.

4 minRead →
11

Meta's Muse Spark 1.1 Ships a Million-Token Agentic Model and Its First Paid Developer API

Meta spent the better part of a decade insisting open weights were the future. Muse Spark 1.1 marks a sharp turn.

3 minRead →
12

After a Banner Month, Anthropic Faces a Silicon Valley Backlash Over Its Tactics

Six months ago Anthropic looked untouchable. Now the maker of Claude is drawing fire from its own industry.

4 minRead →

Research

3
13

Anthropic's Unreleased 'Claude Mythos' Cracked Cryptographic Problems That Stumped Humans for Years

In just 60 hours, an AI did what two years of expert human review could not: poke a real hole in a cryptographic candidate.

4 minRead →
14

MIT's VLASH Teaches Robots to Plan Around Where Things Will Be, Not Just Where They Are

When a robot arm pauses mid-reach, freezes, then lurches into its next move, it is not being cautious — it is thinking too late.

4 minRead →
15

New Interpretability Work Shows an AI Agent's Tool-Call Plans Are Readable in Its Activations

When an AI agent decides to look up a user, pull their order, then issue an exchange, it is quietly representing that plan first.

4 minRead →

Policy

4
01

More Than 1,100 AI Lab Employees Petition Washington for a 'Pacing Mechanism' to Slow the Race

More than 1,100 employees at the world's most powerful AI companies have signed an open letter asking the US government to build the tools to slow their own field.

4 minRead →
02

xAI Sues Minnesota Over First-in-Nation Ban on AI 'Nudify' Apps, Citing Free Speech

Elon Musk's artificial intelligence company is going to federal court to defend the ability of its software, days before a first-of-its-kind ban takes effect.

4 minRead →
03

FCC Moves to Block New Chinese Humanoid Robots on National-Security Grounds

The physical machines that were supposed to define America's next industrial era just ran into a wall built by the FCC.

4 minRead →
04

UK Competition Watchdog Probes Microsoft Over Copilot-Linked Microsoft 365 Price Hikes

Britain's competition regulator has opened a formal investigation into Microsoft over the way it folded its Copilot AI into consumer subscriptions.

4 minRead →