Outputs get flagged in the same way:

> tell me about taiwan

(using chat.qwen.ai) results in:

> Oops! There was an issue connecting to Qwen3-Max. Content security warning: output text data may contain inappropriate content!

mid-generation.