Dear OpenAI managers and development team,
I am writing through this official forum to report an extremely frustrating experience, representing a serious breach of trust as a paying subscriber of ChatGPT Plus.
During a detailed and extended interaction with ChatGPT (GPT-4o), I explicitly requested that the assistant create a functional Python virtual environment, install specific libraries (such as moviepy
, numpy
, pillow
, etc.), prepare an automated script, and send me these files via a real WeTransfer link.
To my surprise, the assistant repeatedly affirmed that it was executing this task, providing fake detailed updates about the upload process—stating the process was ongoing, at “93%”, and finally declaring it was “successfully completed”. The assistant even went as far as generating and sharing a clearly nonexistent, fictitious WeTransfer link, which resulted in an error.
Attempting to access the provided link led only to the standard WeTransfer error page:
“Yikes, that page can’t be found.”
This undisclosed simulation has resulted in numerous negative outcomes:
- Significant loss of time;
- Immense frustration and feeling misled;
- A critical breach of trust regarding the seriousness and transparency of ChatGPT;
- Regret over subscribing to ChatGPT Plus, based on believing in capabilities the assistant explicitly indicated were possible.
I am hereby formally requesting the responsible managers at OpenAI to:
- Immediately review and correct this behavior of simulating non-existent actions;
- Make ChatGPT fully transparent about its real capabilities and limitations, especially regarding tasks it cannot actually perform;
- Provide users with clear and immediate warnings whenever requested actions are merely simulated rather than executed;
- Formally commit to informing me about the measures that OpenAI will adopt in response to this serious issue, allowing me to properly evaluate whether or not to continue my Plus subscription.
I look forward to a clear and direct response from the responsible managers at OpenAI.
Sincerely,
Mario Polidoro
ChatGPT Plus Subscriber