The Scarlet Summer of Our Luddite Discontent: Opening Up on AI Shame & Anthropic Watermarks

“I don’t feel any shame at all recruiting an army of AI agents to help me accomplish my goals. But I think society is trying to make me feel shame.” — Keith Teare   Welcome to the summer [https://www.wsj.com/tech/ai/the-summer-that-america-became-a-nation-of-luddites-5e6dc3e7?mod=hp_lead_pos5] of America’s luddite discontent. Not only are left and right calling for an AI ban, but tech loathing has even infiltrated Silicon Valley in the form of Anthropic’s decision to watermark Claude’s output. That, at least, is the view of That Was The Week [https://www.thatwastheweek.com/] publisher Keith Teare [https://www.thatwastheweek.com/], who believes that watermarking is “nuts.”   In our luddite zeitgeist, Keith believes that a watermark is a scarlet stamp of implied guilt. By marking its own product, Anthropic is not only furnishing the luddite mob with a convenient surveillance tool, but also acknowledging a tacit guilt about using Claude.   “You should never adjust to the market,” the Silicon Valley serial entrepreneur explains. “The market’s wrong nine times out of ten.”   As for Dario Amodei, Claude’s dad (so to speak), Keith is even more accusatory. The Anthropic CEO is kissing ass “like a politician, not the leader of a company,” Keith hisses, repeating this week’s whispers that Anthropic is becoming an “ideologically motivated cult.” If Dario were female, Keith would probably make him a witch. Or, at least, a hussy — the hysterical implication, if not the word, of the Journal’s profile [https://www.wsj.com/tech/ai/claude-dario-amodei-wife-anthropic-e1eeda7d] of Cami Clark, Dario’s wife.   I took Anthropic’s side — somebody had to. I see watermarking as just another AI tool. It’s an honest human fingerprint in an age of fakery. Live by the algorithm, die by it too. Anyway, a watermark has its uses. Rather than moral opprobrium, it feeds our curiosity about who is and isn’t using AI. So, for example, I ran Keith’s editorial through another AI detector and it indicated 86 percent AI-generated while mine came back as 100 percent human. And if I can elude the scarlet stamp, then even I might have to join the luddite mob.   Five Takeaways   •       “Watermarking Is Nuts.” Keith’s editorial takes aim at Anthropic’s decision to watermark Claude’s output — invisible text signals identifying the machine’s hand, cousin to what Google has long done with audio. His objection isn’t the mark itself (“I don’t care that it’s watermarked. I care that Anthropic thinks it should be watermarked”): a watermark starts from suspicion, implies a problem where none exists, and tells Anthropic’s own customers their use of the product needs flagging. Restream doesn’t watermark this show; Adobe doesn’t watermark its exports. To Keith it’s pure zeitgeist capitulation — and while one Claude Max subscriber in the newsletter canceled in protest, Keith hasn’t. Andrew’s rebuttal: watermarking serves curiosity, and in an age of fakery an honest fingerprint is a very human addition. •       The Scarlet Watermark: On AI Shame. The deeper argument was theological. Keith — an atheist who concedes atheism is also a faith, and who believes in good and bad but not sin — feels no shame “recruiting an army of AI agents” so long as he can stand by the output; shame belongs to spam factories and fakers who publish unread. But society, he says, is trying to make him feel it: the canceled author who lost his agent last week proves there’s an issue. Andrew, playing therapist, diagnosed a flicker of guilt anyway — and argued shame is precisely the point: one of the last all-too-human spaces, worth preserving in an age of machines. Both then confessed their methods on air: Pangram scored Keith’s editorial 86 percent AI and Andrew’s 100 percent human. “I’m a clev