Anthropic Is Banning Abuse of Its AI — What Actually Happens When You Hurl Insults at Claude?

This episode looks at a Forbes JAPAN article about verbal abuse directed at AI — and Anthropic's reported move to add a clause to its terms of service prohibiting sustained, unnecessary abuse of Claude.

It touches on something quietly counterintuitive: because generative AI learns from human language, directing insults at it can pull responses in a rougher direction — not because the AI is hurt, but because it has absorbed patterns of how humans respond when they're being insulted. A strange kind of mirror.

There's also a wider question about what this means collectively — whether the way society as a whole interacts with AI might gradually shape something like its character, and a Nietzsche quote from the article asking whether we want to raise AI into something crueler than humans.

The episode draws an observation from working in an optician's shop: that language has a way of setting the tone of a whole conversation, and that harshness, repeated often enough, starts to feel normal.

A quiet reflection on the idea that courtesy might not only be for the other party's sake — but a choice about what kind of language you live inside.