Remember this essay from March?
Many people have come to this view:
You can read the full story – which addresses a wide range of concerns, from business practices to the feeling that Doomerism is overblown to concerns about whether Anthropic is trying to bring competitors to their knees –Here. For many, the tipping point seems to be the events of the last few days, when much of the industry, but not Anthropic, came out forcefully in favor of “open-weight” models (which can be downloaded given sufficient hardware, but are not fully open source because their training sets are generally not publicly available, for example). Many signed a letter; Amodei instead issued a statement justifying his opposition, which many viewed as tone-deaf and self-serving, in part because the statement called for stopping distillation by competitors even though it was just coming out The company destroyed rare books in bulk to distill their contents.
Hedgie@HedgieMarkets
🦔AI companies buy rare books in bulk, scan them with high-speed spine-cutting machines, and destroy the originals. A service called ISBNdb allows orders of up to a million books and ensures that buyers remain anonymous. Pre-2022 books are premium because they are
12:18 am · July 27, 2026 · 22.3 million views
3.96K replies · 27.7K reposts · 70.5k likes
I found myself (somewhat unusually) wholeheartedly agreeing with David Sacks:
David Sacks@DavidSacks
Anthropic claims that it is entitled to free training for all production worldwide, even if the author objects. But if a competitor uses Anthropic’s production after paying for it, it’s IP theft. The hypocrisy is breathtaking.
Be Lehmann @itsolelehmann
Incidentally, Anthropic’s internal document on this topic literally said: “We don’t want it to be known that we’re working on it.” it was called Project Panama. Here’s exactly what happened: 1: Anthropic concluded that books were the cheapest way to build a world-class model because they
10:20 p.m. · July 27, 2026 · 1.47 million views
447 answers · 4.07K reposts · 32.6K likes
All in all, it seems that most of the goodwill that Amodei built up through his opposition to Hegseth is gone, just as most of the goodwill toward Altman eventually disappeared as well.
Perhaps one or the other, Altman or Amodei, will regain the public’s trust, but it is far from obvious that they will. In Altman’s case, I think dishonesty and lack of openness are the core problem; For good reason, people just don’t trust him anymore. For Amodei, who may mean well, I think it’s more about tone-deafness and self-assurance, coupled with a tendency to make arguments that often seem self-serving.
I apologize for the AI’s mistake (the text below is particularly hilarious) and have created a meme in their honor. Coming up: a guy who could sell refrigerators to Eskimos (while probably lying about their future benefits) versus a guy who often seems incapable of reading a room:
The fight for the AI’s son
Seriously, the same WSJ article mentioned above states: “Technologists and policymakers say they are also concerned that AI competition could turn into a duopoly, with Anthropic and OpenAI essentially becoming dominant players controlling the market.”
Considering how harmful AI could ultimately be, we need people we can fully trust, and I don’t think we have any.
PS: I started a survey about X an hour ago. It’s not remotely scientific, of course, but the early results are pretty clear. The good news (for Dario) is that they prefer him to Sam 2:1; The bad news is that the vast majority of respondents don’t trust any of them.
Am I the only one who thinks it’s crazy to spend a few trillion dollars to support a few companies run by people hardly anyone trusts?
https://garymarcus.substack.com/p/dario-takes-it-on-the-chin
