OpenAI’s “too powerful” problem 🔐

What we’ll cover today:

🔐 OpenAI’s Astra triggers tougher safety rules

⚖️ Anthropic beats the Pentagon in court

🎨 Google launches its answer to Canva

👔 How AI Could Build a $1B Company

Anthropic Beat the Pentagon ⚖️

• A federal judge ruled the Pentagon acted illegally when it labeled Anthropic a supply chain risk over limits on military use of Claude.

• Anthropic had refused to allow Claude for autonomous weapons and domestic surveillance.

• The ruling says national security cannot be used as a blank check to punish companies that push back.

OpenAI’s Astra Crossed a Line 🔐

• OpenAI says Astra is its first model powerful enough to trigger its toughest cybersecurity safeguards.

• The model can find and exploit unknown software flaws with very little human help.

• OpenAI will release Astra only to a limited group and monitor it closely.

• The uncomfortable part is that the same skills that make Astra useful for defense also make it useful for attacks.

Google Is Coming for Canva 🎨

• Google is launching Google Pics, an AI design tool for creating visuals mainly through prompts.

• It can make posters, social posts, illustrations, and other work graphics inside Google Workspace.

• Google is putting Pics directly into Docs and Slides, with Drive coming later.

• Canva’s bigger threat may not be better AI design. It may be Google making design a built-in feature people never need to leave Workspace for.

🛠️ TOOLS

🧩 Sider Code – Change how websites work with prompts. →
🎓 Gauth AI Course – Turn any subject into interactive AI courses. →
🎨 Doop – Design with AI agents on shared canvas. →
🎙️ Loqua – Turn voice into writing, coding, and workflows. →

📚 RESOURCE

How AI Could Build a $1B Company 🚀

Replit CEO Amjad Masad explains how AI is changing entrepreneurship, vibe coding, and what one person can build. He also shares how Replit grew, what builders get wrong, and why grit still matters.

See how AI changes startups →