AB Kuai.Dong|Nov 19, 2026 10:24
A small company that was so afraid of AI for 23 years that it couldn't sleep has now become the largest IPO in history. On the eve of Claude's Anthropic IPO, renowned tech journalist Kevin Roose unearthed an in-depth report he wrote while stationed at Anthropic in 2023, which contained many interesting tidbits.
Back then, Anthropic employees compared themselves to J. Robert Oppenheimer, the father of the atomic bomb, and the most popular book in the company was "The Secret History of the Atomic Bomb."
Months before the release of ChatGPT, they had already created a version of Claude, but held off on releasing it for fear of misuse, allowing ChatGPT to become a hit first.
Employees said over lunch that there was a 20% probability that rogue AI would destroy humanity within 10 years. Other employees said they often couldn't sleep because of concerns about AI. In other words, the AI threat theory had already emerged internally three years ago.
Then, when founder Dario was asked about rumors circulating outside—why continue developing AI if he felt it was dangerous—he responded that if people with a conscience didn't participate in AI development, then only those who wanted to do whatever they wanted would be left doing it, which isn't a good thing. This implied that Dario himself has a conscience.
Share To
Timeline
HotFlash
APP
X
Telegram
CopyLink