---
title: "Comment by Josh McCann on \"Demystifying OpenAI's Terms of Use with Regards to Dataset Licenses\" | Hashnode"
description: "I'm very curious how this TOS holds up considering the fact that OpenAI likely broke many TOS in order to train their models. I can't imagine it would be possib"
url: "https://hashnode.com/posts/demystifying-openais-terms-of-use-with-regards-to-dataset-licenses/67251e152aaf46e70081e0ae/comment/677ae2767b306072125912fe"
---

[Discussion](https://hashnode.com/posts/demystifying-openais-terms-of-use-with-regards-to-dataset-licenses/67251e152aaf46e70081e0ae)

Comment [Demystifying OpenAI's Terms of Use with Regards to Dataset Licenses](https://hashnode.com/posts/demystifying-openais-terms-of-use-with-regards-to-dataset-licenses/67251e152aaf46e70081e0ae)

[Josh McCann](https://hashnode.com/@ArtisticMusician)

[Jan 5, 2025](https://hashnode.com/posts/demystifying-openais-terms-of-use-with-regards-to-dataset-licenses/67251e152aaf46e70081e0ae/comment/677ae2767b306072125912fe)

I'm very curious how this TOS holds up considering the fact that OpenAI likely broke many TOS in order to train their models. I can't imagine it would be possible to train their models without breaking a number of TOS. The vast amount of data used just seems impossible they could make that many deals already to cover it all. Maybe I'm wrong and it's totally possible.
