Can you get banned for insulting Claude? Here’s what policy says

Call Claude useless after a botched answer, and you're fine. Keep insulting it for no reason at all, and you may find it simply stops replying. That's the line Anthropic has drawn in an update to its usage policy.

Jane Manchun Wong first flagged the change on X, sharing a screenshot of the new wording. It rules out repeatedly directing unnecessary abuse or cruelty at Anthropic's AI models, Claude included.

What Anthropic usage policy on Claude abuse does it cover?

The target is narrow. It's extreme, repeated cruelty with no real purpose, not a bad day with a chatbot. You can still disagree with Claude, point out its mistakes and tell it exactly how badly it did.

Legitimate testing isn't the aim either. The policy separates people who want to use, challenge or probe the model from people who engage just to be abusive.

Edwin at Claudedevs responded to Wong’s post, saying that Claude’s ability to stop interactions when abuse happens is still the most important thing here. So the outcome will probably be that Claude will no longer respond to you, and not that Anthropic will do something to your account.

There is no explicit mention of banning in the available wording. Do not take that as a promise that there will be no other penalty in Anthropic's policy regarding accounts.