Anthropic pays $1.5B for "borrowing" books

💸 Anthropic Writes a $1.5B Check for "Accidentally" Using Pirated Books

• Anthropic just settled a copyright lawsuit for $1.5 billion after getting caught training Claude on what's basically the Library of Alexandria... if it were hosted on sketchy torrent sites.

• The "move fast and scrape everything" era just got expensive. Every AI company is now doing the math on whether paying authors upfront is cheaper than paying lawyers later. (source)

---

🇪🇺 ASML Drops $1.5B on French AI Startup, Europe Declares Independence

• Chip giant ASML just became Mistral AI's sugar daddy with a $1.5 billion investment, making the French startup worth $11.7 billion and officially Europe's AI golden child.

• Europe's tired of watching Silicon Valley and China play AI chess while they make croissants. This is their "we're building our own ChatGPT, with blackjack and GDPR compliance." (source)

---

🤥 OpenAI Finally Admits Their Models Are Pathological Liars

• OpenAI published a paper explaining why AI hallucinates: because current training rewards confident BS over honest "I don't know" responses. It's like grading students who guess on every question higher than those who admit ignorance.

• The biggest AI lab just said "our evaluation methods are fundamentally broken." That's like McDonald's admitting their ice cream machine is always broken — except this affects global information. (source)

---

🤖 Motion Raises $38M to Build "Microsoft Office for Robots"

• AI agent startup Motion just raised $38M to create a suite of business bots that handle scheduling, sales, and customer support. They hit $10M revenue in four months, which is either impressive or terrifying.

• We're speedrunning from "AI can write emails" to "AI can replace your entire entry-level workforce." Your intern might be a Python script by Christmas. (source)

---

⚖️ AI Forces Lawyers to Question the Sacred Billable Hour

• Law firms are having an existential crisis: if AI does the grunt work in minutes instead of hours, do they bill clients for robot time or find new ways to justify their Porsche payments?

• When lawyers start questioning how to make money, you know AI disruption is real. Next up: consultants wondering if PowerPoint automation threatens their $500/hour slide fees. (source)

---

📱 InDrive Wants to Be the Everything App (Again)

• The world's second-biggest ride app is expanding into grocery delivery in Kazakhstan, because apparently every tech company's endgame is becoming WeChat for their region.

• Another "super app" enters the chat. Soon you'll need 47 different apps that all do the same 12 things, but with slightly different logos. (source)

---

🎵 Amazon Music Launches AI Playlists to Fight Spotify

• Amazon's new "Weekly Vibe" feature uses AI to create personalized playlists every Monday, because apparently the streaming wars needed more algorithms deciding what you should feel.

• The battle for your eardrums is now fought with machine learning. Your music taste is just training data for the next quarterly earnings call. (source)

---

🔍 Spotify Adds Smart Filters Because Your Library is a Mess

• Spotify Premium users can now sort their music by mood and activity, because having 10,000 liked songs without organization is the digital equivalent of hoarding.

• We've created so much digital stuff that we need AI just to find our own things. Peak modern problems require peak modern solutions. (source)

---

Until tomorrow (if the lawyers don't shut us down),
Team Galaxy.ai

P.S. If you're an author, maybe check if your book is training the next ChatGPT. Could be worth $1.5 billion, apparently.

P.P.S. To the law firms reading this: we're not legal advice, we're just sarcastic commentary. Please don't bill us for reading this newsletter.