AI labs are scanning books to train AI
Amazon winds down its own AI
Nvidia bets on former OpenAI Chief Scientist’s new AI lab
Build AI Human Agents that Support, and Scale Your Business For You
Big companies like Amazon, Salesforce, and Deloitte are putting AI video agents on their websites. Now you can too, without a developer and a huge budget. A PAL is an AI video agent you can build in Tavus PAL Maker: it can see, hear, and talk with your customers face-to-face in real time. Describe what you want your PAL to do in a few sentences, say, “a friendly front-desk helper that books appointments and answers FAQs,” and Tavus will build it for you. Deploy it to your site in a few clicks. It handles questions 24/7, speaks 50+ languages, and remembers repeat visitors. You can even upload a custom face to build brand awareness. Build a PAL today and see what your storefront’s been missing. Build a human edge for your brand, your customers will thank you. Use code SBA for 50% off!
AI labs are scanning books to train AI
HAARLEM, Netherlands — It was an ordinary summer afternoon when an unusual email landed in Pieter de Vries’ inbox.
The Dutch antiquarian bookseller didn’t think much of it at first. A woman named Natalia, writing on behalf of a company called “2077AI,” said she was looking to place a “fairly large order” of books. Attached was a spreadsheet listing more than 3,000 ISBNs, along with instructions to match titles, prepare quotes, and estimate shipping costs — to China.
De Vries barely glanced at it. He assumed it was spam.
It wasn’t until weeks later, when a Dutch journalist investigating the procurement request reached out, that he learned the order was tied to artificial intelligence. “I was shocked,” de Vries said, recalling his reaction. He provided Fortune with the original email and the accompanying spreadsheet, which listed 3,001 titles — most published between 2020 and 2021 by academic presses including Emerald Publishing, Elsevier, Wiley, Routledge, and Oxford University Press. The subject matter ranged from business and education to engineering, public policy, and medicine.
“Very unusual,” de Vries said. “That’s why I regarded the request as spam.”
The episode is the latest sign of how AI companies are looking beyond the open internet for high-quality text to train large language models — increasingly turning to physical books as a source of clean, well-edited data. Scrutiny of the practice intensified last summer, when court records revealed that Anthropic had purchased millions of physical books, cut off their bindings, scanned the pages, and discarded the originals to build a searchable digital library for training its AI models — an effort internal documents referred to as “Project Panama,” according to The Washington Post. The technique, known as “destructive scanning,” strips a book’s spine so its pages can run through high-speed scanners before the remaining physical copy is thrown away....
Amazon winds down its own AI
Amazon.com is retooling its artificial intelligence strategy, winding down several in-house AI models as it pivots toward a new frontier-model initiative, Business Insider reported Tuesday, citing people familiar with the matter.
The company has begun phasing out most of its flagship Nova models, including the high-end Premier and Omni models, the Reel video-generation tool, and the Canvas image-generation tool, according to the report.
In their place, Amazon is redirecting resources toward a frontier-model effort led by researcher Pieter Abbeel, with frontier research emerging as a top company priority this year, the report said. The new flagship model is expected to debut at Amazon’s re:Invent conference later this year, Business Insider reported.
The shift comes as Amazon has struggled to match the buzz generated by rivals OpenAI, Anthropic, and Google around their respective AI models. Instead, the company has leaned into its role as a cloud and infrastructure provider, positioning itself as the backbone enterprises rely on to build and run their own AI systems.....
Nvidia bets on former OpenAI Chief Scientist’s new AI lab
Nvidia is investing in Safe Superintelligence, the secretive AI lab founded by former OpenAI chief scientist Ilya Sutskever, deepening a partnership that could give the startup greater access to computing power while helping the chip giant guard against rival hardware makers gaining ground.
The chipmaker made what the companies described as a “substantial” investment after receiving a rare look inside the startup’s research, according to a joint statement Monday. As part of the deal, Safe Superintelligence will gain access to vast quantities of Nvidia’s flagship graphics-processing units — enough to boost its computing capacity by an “order of magnitude,” the companies said.
It’s the second such deal Nvidia has struck with a prominent OpenAI alum. In March, the company invested in Thinking Machines Lab, the startup co-founded by former OpenAI Chief Technology Officer Mira Murati. That lab released its first AI model, trained on Nvidia hardware, earlier this month.
Sutskever, a researcher widely credited with helping spark the modern AI boom, co-founded Safe Superintelligence in 2024. The company says it has one goal: a “straight-shot” effort to build what Sutskever calls “safe superintelligence” — AI capable of matching human performance across the board without posing a threat to humanity. Its research is “focused on overlooked aspects of how the human brain functions,” Sutskever said in a statement.
In 2012, Sutskever co-authored a landmark paper with Geoffrey Hinton, the Nobel Prize-winning AI researcher, on computer vision. That work helped ignite the modern AI race by demonstrating that scaling deep neural networks with massive amounts of data and computing power could yield measurable gains in machine intelligence….
Read the original on Wall Street Journal