Grant Harvey

From the developer wars between OpenAI and Anthropic, to $690B in datacenter spending, to ARC-AGI benchmarks getting demolished, to the OpenClaw fork explosion... we tried to capture everything happening in AI in a single article.

The Supreme Court says corporations can speak. It says they can pray. So could a company refuse to go to war on religious or moral objections? The Pentagon's beef with Anthropic create an interesting thought experiment.

Anthropic refused to let the Pentagon use its AI without limits. Within 48 hours, it was blacklisted, replaced, and still being used in combat. Here's everything that happened.

From Anthropic's Pentagon standoff to Nvidia's Groq-powered chip to the AI scare that tanked markets, here's every story we tracked this week.

Anthropic released a comprehensive guide to building "skills" for Claude, reusable instruction sets that let you teach it your workflows once and benefit every time.

Google launched Nano Banana 2, merging Pro-level quality with Flash speed. It's now the default across Gemini, Search, Flow, and Ads in 141 countries

Perplexity just launched "Computer," a system that orchestrates 19 specialized AI models to handle entire projects end-to-end. Here's what it does, how it works, and why it matters.

Anthropic launched 10 new Cowork plugins that turn Claude into a specialized agent for HR, design, engineering, finance, and more, plus cross-app workflows between Excel and PowerPoint.

Inception's Mercury 2 and Logical Intelligence's Kona both reject the token-by-token approach powering ChatGPT, Claude, and Gemini. One is 10X faster. The other is 50X more accurate on reasoning. Here's why it matters.