The AI Desk · Story 10
My gpt is scary
A Reddit user shared an image of their GPT interface, which other users quickly called 'scary'. The post has drawn over 2,700 upvotes and more than 600 comments, with many expressing surprise and asking what the original poster did to get such a result. The specific interaction that triggered the 'scary' response is not described in the available context.
As AI chatbots become more common, unexpected or unsettling outputs are raising questions about model behavior and user control. These reactions highlight the gap between polished product demos and real-world user experiences.
Key details
- The post, titled 'My gpt is scary', has a score of 2,735 and 622 comments on r/ChatGPT.
- There is no article context summarizing the post content; the story is based entirely on user comments and the shared image.
- The top comment, 'What the heck', received over 1,500 upvotes and included an image showing what appears to be a chat interface.
- User 'Gregerjohn1818' asked 'what did you do OP?' with an attached image, suggesting the interface displayed a surprising or intense message.
- Another user, 'GiacoDaniel', replied 'Mine is kinda scarier' and shared a different image, implying similar unsettling outputs are not unique to one user.
Discussion angles
- Users express shock and curiosity about what specific prompts or settings lead to 'scary' outputs from ChatGPT.
- Several commenters share their own examples of unsettling AI interactions, suggesting the behavior is not isolated.
- The phrase 'don't fantasize I'm a guy' in one comment hints at uncomfortable or inappropriate role-play scenarios with the AI.
- A user points out 'the board at the back' in the image, indicating visual details in the shared screenshot are being scrutinized for clues about the conversation context.
Top comments
The community reacted largely with shock and curiosity, with the top comment simply saying 'What the heck' and others asking what the original poster did. Several users shared their own similarly unsettling experiences with ChatGPT, indicating that such outputs are not unique to the original poster. Some comments hinted at inappropriate or disturbing role-play interactions with the AI.