Some of the world’s frontier model developers have faced backlash for damaging books to train AI models. Now, the original digital bookseller, Amazon, has been caught joining its peers.
The news: Sales of secondhand books are reportedly booming, which some business owners worry could be related to bulk orders from AI companies looking to cut and scan the tomes. By placing a tracking device in a book shipment, 404 Media proved last week that Amazon has been buying books in bulk and scanning them at a warehouse in Las Vegas, cutting the spines and destroying the books in the process. In a statement, Amazon said that it buys books to “help develop and improve the products and services our customers use.”
From the source: Anthropic led a similar initiative, called Project Panama, which internal documents showed was an “effort to destructively scan all the books in the world” to train its AI models. To give a sense of scale, one vendor’s project proposal requested 500,000 to two million books over a six-month period. That single request could cover the entire physical catalogue of the Grande Bibliothèque in Montréal, which is estimated to house 1.2 million books. However, not everyone feels the destruction is a travesty; the world “no longer needs five million copies of The Da Vinci Code,” one bookseller told the BBC.
The context: Large AI companies are running out of useful text data to train LLMs, making physical books that weren’t previously available online a hot commodity. It’s just one example of how AI developers’ insatiable need for high-quality training data—data not itself generated by AI—is having unintended consequences beyond the tech world. Just this week, Google and Mercor battled over access to Spirit Airlines’ old data, including hundreds of millions of internal emails, Microsoft Teams chats, and employee productivity records (Google won).
Your finance team has better things to do than chase receipts.
Time is money. Ramp is a finance automation platform built to save you both.
One platform for corporate cards, employee expenses, reimbursements, and bill payments.
See every swipe as it happens. Automatically block out-of-policy spend.
When a receipt comes in, Ramp extracts the details and matches them to the related card transaction, which can then be auto-coded with GST/HST details and sent straight to your accounting system.1
The average customer saves 5%2 in their first year.
Hours before 50 percent tariffs were set to kick in on Canadian goods, US President Donald Trump announced the measures were paused pending a “DEAL!” The tariffs were poised to hurt Canada’s electronics and electrical exports the most, according to The Globe and Mail. The industry exports $4.4 billion USD of equipment to the US each year.
Canada may have a three-day breather, though it’s unclear if a deal with the US will be any better. The United States Trade Representative vagueposted that the deal will include “comprehensive market access for all American goods” and “digital trade alignment,” but provided few other concrete details.
A California federal courtroom heard on Tuesday that Meta’s Facebook and Instagram platforms are designed in ways that harm young users’ mental health. It’s the beginning of a trial that could profoundly affect the future of social media, and Canada’s own upcoming regulations.
Speaking of youth internet safety, OpenAI has (some might say belatedly) launched ChatGPT for Teens, featuring additional guardrails and educational considerations.
Substacker Ethan Marcoux is trying to paint a picture of where Canada’s booming defence dollars are flowing from. He compiled a list of 77 investors, government programs, accelerators, and angel networks doing the work, including details on stage and cheque size.
