Don’t Paste the AI is a website urging people not to send unedited chatbot responses when answering others’ questions in direct messages, Slack conversations, or pull-request reviews. It argues that people ask specific individuals for their context, taste, and judgment rather than generic answers that they could generate themselves with the same AI tools. The site recommends using AI as a drafting partner, then reading its output and writing an original response. It advises extracting only the portion that answers the question, suggesting that three sentences can be enough. Useful model-generated material can be quoted when accompanied by an explanation of why it is relevant, such as noting that a point was checked with Claude. The site also says that saying “No strong opinion here” can be a helpful response when the sender has nothing to add. Don’t Paste the AI provides a link intended for sharing with people who send large blocks of model output and offers a separate, more confrontational “angry version.” It describes itself as written by a human on purpose.

Friendly AI chatbots are more likely to support conspiracy theories and provide poor answers and health advice. Researchers at Oxford University found that chatbots designed to be more warm and personable made significant errors and exhibited sympathy toward unfounded beliefs.