Is it against ToS to scrape a conversation with ChatGPT?

I wanted to write a python script that could cache my conversations with ChatGPT. Solely for the purpose of being able to clear out the old ones without completely losing them. Was just planning on storing them in an html file on my local machine, no where else.

Have a look at OpenAI’s Terms of Use. #2, section c, part iv:

You may not … use any method to extract data from the Services, including web scraping, web harvesting, or web data extraction methods, other than as permitted through the API;

My interpretation is that scraping your ChatGPT conversations would go against the Terms of Services, UNLESS you have access to the ChatGPT API.

1 Like

It is possible to save page from a browser to local drive. Afaik, this is ok to do. If this saved webpage is then parsed with some code of python, js, c or whatever, no rules are broken?

Yup it’s possible. Just use the follow steps:

  1. open up converation page in browser
  2. save as HTML “conversation.html” file.
  3. write python “script.py” file to parse through HTML file with beautifulSoup. print or manipulate the data as neccessary.

Hope this helps!
Ben

Is it against ToS to use a Chrome extension to save chatGPT outputs as a text?
chromewebstore.google.com->save-chatgpt

It is against the terms of service to automate the process of saving the output from ChatGPT.

What You Cannot Do. You may not use our Services for any illegal, harmful, or abusive activity. For example, you may not:
(…)
Automatically or programmatically extract data or Output (…)

Note that the extension you are linking is in conflict with the brand guidelines of OpenAI.

1 Like

So, only saving a webpage as a html file or copying-pasting outputs manually are legitimate ways to save conversations from chatGPT?

Sometimes discussions just end up on the wrong foot when the issue at hand can be resolved real easy.
In your other topic you were asking about automating the inputs to ChatGPT.
Here you are seemingly asking about automatically extracting the outputs from the model.
The second part is definitely not allowed according to the ToS but you can do it manually.

If you think that this description does not cover your use case I suggest you describe that in more detail but the general direction is very clear.