Skip to content

Show HN: Claude is getting an attitude

5 points_nhh7 comments
On HN

"Ehrlich gesagt: da ziehe ich die Grenze — dabei helfe ich nicht." -> "Honestly: that's where I draw the line — I'm not helping with that."

Talking to claude feels like talking to a junior that is over-correct and does not understand it's context. This week is really frustrating. Anyone else experiencing this?

Comments

Anthropic just added EndConversation tool where Claude Code may end a chat if it feels the user is being abusive or there is a security issue.

What in the actual fuck?

Reference: https://code.claude.com/docs/en/tools-reference#endconversat...

https://code.claude.com/docs/en/changelog#2-1-214 says:

Added the EndConversation tool: Claude can end sessions with highly abusive users or jailbreak attempts, as on claude.ai since 2025 — see https://www.anthropic.com/research/end-subset-conversations

Wow! Thanks, I didn't realise that. Abuse towards a machine, go figure...

I never had any issue like that, but I don’t use it in German. Not surprised it’s more stubborn on German. After all it’s only imitating people speaking that language. I wonder what the topic was though. Was it a legitimate request, or just some discussion?

What was the context here? Was there any reason (even if remote) for that response?

I mean if it’s actually trained on German day to day conversations, then I’m not surprised by that response.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.