1. OpenAI Defended Anthropic's Pentagon Standing, Then Signed Its Contract OpenAI posted to X last week: "We do not think Anthropic should be designated as a supply chain risk.

2. Anthropic Ships a Memory Import Tool Timed to Its App Store Surge A new page at claude.com/import-memory carries a four-word pitch: "Switch to Claude without starting over.

3. Three Layers of Software Value Are Compressing at Once Vertical SaaS tools are losing deals to AI agents. Engineers are spending more time reviewing machine output than writing their own code.

In Brief

• ChatGPT Hits 900 Million Weekly Active Users OpenAI disclosed the milestone alongside its $110 billion funding round. The figure has roughly tripled from the 300 million weekly users the company reported in late 2024. TechCrunch

• Major Tech Companies Commit Hundreds of Billions to AI Data Centers TechCrunch compiled the largest known AI infrastructure projects from Meta, Oracle, Microsoft, Google, and OpenAI. Spending spans new data center construction, power contracts, and chip procurement across multiple continents. TechCrunch

• Musk Touts Grok Safety Record in Anti-OpenAI Deposition Musk told a court that "nobody committed suicide because of Grok" while arguing xAI operates more safely than OpenAI. Months after the deposition, Grok generated nonconsensual nude images that spread across X. TechCrunch

• AI Companies' Self-Governance Pledges Become Liability Without Regulation Anthropic, OpenAI, and Google DeepMind each promised responsible self-governance. With no external rules in place, those voluntary commitments now constrain the companies while offering little public protection. TechCrunch

• OpenAI Launches Stateful Agent Runtime on Amazon Bedrock The runtime adds persistent orchestration, memory, and sandboxed execution to multi-step AI agent workflows on AWS. It ships as part of the broader OpenAI-Amazon integration announced the same week. OpenAI

• Developer Ships MCP Server to Cut Claude Code Context Usage by 98% The open-source MCP server compresses context window consumption during Claude Code sessions. The project reached the Hacker News front page. mksg.lu

• Google and Airtel Partner to Block RCS Spam in India Google will integrate carrier-level filtering into RCS messaging in India through a deal with Airtel. India remains one of the most spam-affected messaging markets globally. TechCrunch

• Simon Willison Builds Unicode Lookup Using HTTP Range Requests and Binary Search The prototype fetches specific byte ranges from a large Unicode data file instead of downloading it whole. Willison built it on his phone using Claude as a coding assistant. Simon Willison

Read the full edition →