GPT-4 leaks others questions and answers while "continues >>" with large chunk code production

[lament mica]

Steps to reproduce:
Prompt: Optimize this Arduino code to make it more efficient for power management and speed (than I paste a long 500 lines of Arduino code)

Expected result:
Revised code but its long so expected to be broken into two or three separate answer from GPT-4

Actual result:
It goes fine but once I [ress ":continue>>" button to cvarryu on code production, it injects seemingly random questions and answers from other users (I never asked such questions). Screenshot provides example.

Additional information

Browser: Chrom 114
OS: Mac OS 13.4.1

[spiral sparrow]

This is a hallucination, not another user's request. We likely have an issue w/ the request and the model didn't get the correct context. Can you share your chat so I can take a look?

[lament mica]

Sure, understand. I've extracted all the prompts and descriptive answers plus commented where the code was. I can't share the entire chat.

[spiral sparrow]

can you DM me the email you use with chatgpt?

[lament mica]

Sure, I've send e-mail via DM