📚 Amazon pulps rare books and Cursor fights GitHub

⚡ aiBytes_ #6

Good morning, builders 👋

Welcome back to your Monday-morning dose of aiBytes_. I did not expect to spend a week in AI thinking about paper. But Amazon, the company that started out selling books, is pulping rare ones to feed its models, and Anna's Archive is racing to scan whatever survives 📚

The week the readers pushed back: the labs kept shipping, but the loudest posts of the week were people refusing to read model output - just as a study put a third of the web's new pages down as machine-written.

Today: the book-pulping saga, Cursor coming for GitHub's lunch, an agent that can clip your home videos, seven repos led by a literal money printer, and five essays from the week HN stopped reading.

📚 Amazon, which started off selling books, is destroying rare texts to train AI - rare print is the one corpus the models have not already eaten, so it is going into the shredder. Story of the week (TechCrunch)

📖 AI companies destroy physical books - let's scan rare books before it's too late - Anna's Archive answers with a public call to digitize what is left before it is pulp. (Anna's Archive)

🎭 Israel creates fake think tank in likely attempt to dupe AI chatbots - a content farm aimed not at readers but at the models readers ask. (Responsible Statecraft)

🐙 Cursor capitalizes on GitHub frustration, launches rival hosting platform - your editor vendor would now like your repos too. (TechCrunch)

💸 GPT-5.6 Sol pricing cut by 50% on OpenRouter - the fast tier just got half price, and the margin debate reopened on cue. (OpenRouter)

🚨 AI-generated Copilot "Autofix" allowed compromise of Snowflake's Jira - Wiz walks through how an unreviewed machine patch became an entry point. (Wiz)

👁️ DeepSeek-v4-flash-vision-exp - DeepSeek's experimental fast vision model lands quietly in the API docs. (DeepSeek)