Navigating Copyright Concerns in the Age of GPT: What You Need to Know

I’m a member of the Authors Guild, and I received an email the other day. Here is a section of it:

“As you may know, the Authors Guild is actively engaged in advocacy around generative AI technologies (AI that can generate new text, visual, and audio works—e.g., ChatGPT, MidJourney, Stability—from training on existing works, such as books, photos, and visual art). Our goal is to ensure that appropriate guardrails are put in place around the development and use of generative AI and to protect the robustness, diversity, and very future of the writing profession.

Among other things, we are participating in a Copyright Office listening session on artificial intelligence and literary works on April 19. The questions the Copyright Office has posed for the listening session revolve around writers’ use of generative AI and potential impacts on the profession and the book industry.”

At the end of the email, there was a survey which I participated in.

GPT cannot have access to published books directly

For example ChatGPT was trained on a massive amount of text data from the internet. It was trained on a dataset called the Common Crawl, which is a publicly available web crawl dataset that contains billions of web pages.

The Common Crawl includes a wide range of content from various sources, such as news articles, academic papers, blog posts, and social media posts. In addition to the Common Crawl, ChatGPT may also have been trained on other datasets or corpora depending on the specific version and configuration of the model.

So what does that mean?

If your e-book is publicly available in its entire content, then GPT might have access to it. However, the material will not be added to the dataset if it’s closed content, like what Kindle offers.

That said, if someone copies your book legally or illegally, publishing it on a website for anyone to access it, then the Common Crawl rule applies, and GPT can fetch every word from it.

To sum it up:

Everything, that is all you can eat for search engines, bots, and crawlers GPT can and will add to its dataset to learn from and use the text and art. That by itself will probably re-define copyright laws and the entire SEO itself.