Unexpected internal drafting text appeared in a ChatGPT response — has anyone else experienced this? -GPT 5.6 Sol model
While discussing Oracle PL/SQL security with ChatGPT, I received a response containing what appeared to be internal drafting or response-planning text. It included unfinished phrases, repeated words and notes about how the answer should be written.
When I asked ChatGPT about it, it described the text as an accidental drafting fragment that was not intended to appear in the final response. However, it could not confirm the exact technical reason or how frequently this happens.
I would like to ask the community:
-
Has anyone else experienced something similar?
-
How often have you noticed drafting notes or internal-looking text in a response?
-
Is this generally just a response-generation or formatting error?
-
Could it indicate any privacy or security concern?
-
Is there any possibility that such text could contain information from another user, conversation or private source?
-
Has OpenAI officially documented or addressed this type of issue?
below is the screenshot of the response
