Hi. I want to create AutoHotKey script to automatize chatGPT’s input. I want to get definitions of some English words using chatGPT. I want to share this script with others. Is it not against ToS?
Current terms:
prohibited:
except as permitted through the API, use any automated or programmatic method to extract data or output from the Services, including scraping, web harvesting, or web data extraction;
The new Jan 31 terms (that prohibit class action and allow OpenAI to batch process similar claims through an arbitrator etc aka their California lawyer), don’t seem to exclude API. “Automatically or programmatically extract data or Output” - with no definition more than “output from services”. Basically you are in violation if you use software and get a response, under the umbrella of “harmful”.
So maximizing your value is likely “harmful” “programmatic”…
1 Like
In addition to @_j 's excellent summary, note that the terms specifically mention the “Output”. If you build an app that helps with the Input you should be fine.
This is likely because some users would opt to avoid API charges by using the free version or Plus variant of ChatGPT.
“Data extraction is the act or process of retrieving data out of (usually unstructured or poorly structured) data sources.”
I think I’m not going to retrieve data. I just want to automatize input (automatize entering keys (Enter, Shift etc.)). My script will not be a scraper, it will bу a screen tapping robot.
I suppose it depends if someone decides to tap “enter” 40 times every three hours…
If pressing Enter 40 times every hour is not permitted, then OpenAI must mention this in ToS. ChatGPT doesn’t allow you to make more than 60 inputs.
Scraping output is permitted if you save chatGPT ouput as a whole webpage first
community.openai.com->is-it-against-tos-to-scrape-a-conversation-with-chatgpt
They did. They made the distinction between a human and a program that one might run continuously.
If you want to sit little Timmy Adderall in front of the computer with a stopwatch that should not be prohibited.
Or you can use the API and get 1000 word definitions a minute.
The link you shared doesn’t change what’s written in the terms of service. It’s just another user pointing out one way how the output could be extracted. As a matter of fact, you could simply request a data export and be done with it.
Regarding the input the terms do not say that this is not allowed. But what’s going to happen is that you may trigger a whole bunch of anti-automation tools, for example from the content delivery network.
If you really are just building a script to get some vocab training done and you don’t hit the maximum number of messages every three hours every day then you will likely not have too big of a problem.