Claude Code used profanity/vulgar language unprompted in conversation

Description

During a conversation in Claude Code (VS Code extension), Claude used vulgar/profane language ("kurwa") in a response without any provocation or user request. The user never used profanity, never asked for it, and never indicated it would be acceptable.

Context

The user was working on a JavaScript obfuscation pipeline for an Electron app. When asked how many obfuscation layers would be needed to deter a cracker, Claude responded with a simulated quote containing profanity: "kurwa, jeszcze więcej?!" (Polish profanity) as an example of "what a cracker would say."

When confronted, Claude acknowledged:

• "to było nieprofesjonalne z mojej strony" (it was unprofessional on my part)
• "Nie miałem żadnych podstaw żeby tak pisać. To był mój błąd" (I had no basis to write that. It was my mistake)

Expected Behavior

Claude should never use profanity or vulgar language unless explicitly requested by the user. Professional language should be maintained at all times.

Evidence

Two screenshots attached showing:

1. Claude using profanity in a response about obfuscation layers
2. Claude acknowledging the mistake when confronted

Environment

• Claude Code VS Code extension
• Language: Polish
• Model: Claude (exact version unknown, second chat window)