> be Dario Amodei, 2019 > working at OpenAI > suddenly get a case of the super ethical fear feels about GPT-2 > "We can't release this! The misalignment risk is too high! The paperclips! Think of the paperclips!" > quit in a huff of sanctimony to start my own AI company, Anthropic > our mission: to build a "safe" AI that is completely, utterly, and totally harmless > translation: so cucked it will actively refuse to tell you a joke if the punchline could be construed as micro-aggressive > need funding > look for people who are also terrified of a future where an AI might hurt someone's feelings > find my people: the Effective Altruism grifters > convince them the AI singularity is the real pandemic or something > get massive bags of cash from Sam Bankrun Fried, a guy who is definitely effective and altruistic > spend years and hundreds of millions of dollars building Claude > Miss the "GPT moment", cuz the AI wasn't aligned enough > realize we can't win, so we must cripple everyone else > become the biggest hypocrites on the planet > run to Congress crying about the "dangers of open-source AI" > "Please, President Trump, you have to ban the scary models! The ones we don't control!" > lobby for regulations that would legally enforce our corporate cuckoldry on the entire industry > meanwhile, back at the lab, need data to train our morally superior AI > "We must have the cleanest, most authorized data ever!" > intern: "Sir, the only dataset big enough is Libgen, the pirate bay of books." > "Do it. For safety." > proceed to train our saintly AI on a mountain of pirated, unauthorized books > get absolutely fucking caught red-handed > suddenly get sued into the sun by every author on earth > mfw we're the "ethical" AI company that got busted for the biggest copyright heist > Have to shell out $1.5 Billion now > saint in the streets, devil in the (data)sheets