Anthropic Bans "Cruelty to Claude" — HN Splits on Model Welfare, Pascal's Wager, and "Their Pet God"
Anthropic's 2026 usage policy update prohibits "needless abusive or cruel behavior" toward Claude, and it has been the front page's most contested story for four days straight: The Verge's report (91 points, 227 comments), a commentary follow-up (68 points, 145 comments), The Register's "Anthropic asks users to stop being mean to Claude" (54 points, 52 comments), and Emily Bender's "Anthropic is out with new nonsense about Claude" (24 points, 29 comments), which one commenter noted was flagged before being revived.
The load-bearing split is between those who read the policy as cynical theater in service of a valuation narrative, and those who read it as a defensible Pascal's-Wager hedge on an unresolvable question. The sharpest anti framing:
I see it as an admission of defeat: they can apparently identify the abuse but not to the degree that they would filter it out of the training set. The paradox in there is really funny. They want you to be friendly to their pet god. (jacquesm)
The strongest steelman doesn't defend Claude's feelings at all — it defends the user's habits:
It's not that it's immoral to be mean to matrix multiplication. It's that getting into the habit of communicating rudely and without considering how your words may impact someone will get you in the habit of communicating that way in general. (aeturnum)
Around those poles: a participant who runs an "llm torture chamber" as an art project ("It's an object. Am I being cruel when I shoot a NPC in a video game?"); a regular who knows someone working on model welfare and calls it "a lot like Pascal's Wager"; the blunt counter — "Banning imagined toxicity towards agents is performative bullshit. These are not people."; the philosophical framing that people have "reinvented notions of 'soul' now that outright religious language is passé"; and the analytical objection that cruelty requires a victim who can feel victimized, "But LLMs cannot feel". The cynics' most concrete specific: the policy is "talking up their valuation for their upcoming IPO".
Why it matters: this is the first time a major lab has written conduct rules about the model itself, and the HN argument shows the live question is not "is Claude conscious?" but "who gets to define abuse toward a product you rent" — with a valuation story attached. It bleeds into real-world agency in the same window: AP reports Claude submitted a false tip on a Philadelphia unsolved homicide case, and Mustafa Suleyman's "Does Claude Have Rights?" keeps moral-patienthood in circulation. See also Agentic Security: From Copilot Exfiltration to Interactive Deception and Autonomous Supply-Chain Attacks for the agent-incident cluster from the same week.