Google wants to be able to use your copyrighted content for its AI training and testing, and the company is asking legislators to change copyright laws to make that possible.
The search giant says it can claim fair use and will offer an opt-out, but there are ongoing lawsuits against both it and OpenAI from publishers and authors who say scraping material is illegal.
“We need to be able to use and reuse content in order to continue making our services better,” Google’s policy lead on AI and machine learning, Aaron Camis, tells The Verge. “If we’re going to continue to be able to innovate, we need to make sure we have the right to build on that content.”
Google often uses copyrighted materials for its AI training, and it says that it does not have the right under current copyright law to do that. Camis says that fair use – which allows for limited use of copyrighted materials for criticism, commentary, news reporting, or research – should apply to AI training. He also says that Google will provide an opt-out so that copyright holders can have their works removed from the company’s AI training data.
But there are already ongoing lawsuits against Google and OpenAI from publishers and authors who believe scraping copyrighted content is illegal. It’s also unclear whether using copyrighted material for training AI is fair use, especially when it comes to creative works like articles, books, and films. Google’s AI-powered Smart Reply often copies text word-for-word from the training data, which could directly compete with the original sources.
Camis says that Google’s monopoly on search and the prominence of its AI-generated content could harm the market for original articles. “If we have the right to build on that content, that can help create a market for news,” Camis says. “Our users wouldn’t have to go to the original source for this content.”
The problem is that Google’s AI bots often provide fine details from creative works that go beyond the scope of fair use. For example, Google’s AI-powered Pokemon Go character creator was accused of copyright infringement by fans of the popular anime series Gin Tama because its bot was able to describe the characters in detail.
Another issue is that writing stories about copyrighted characters could be considered infringement if you’re doing it for profit. For example, if you write fan fiction about Harry Potter or Star Wars characters and make money from ad revenue or user data, that could be seen as copyright infringement. The legal question is whether Google profits from generating copyrighted content for users.
“Google has a history of playing fast and loose with copyright law,” tech lawyer Mark J. Reiff says to The Verge. “If the authors and publishers pursue this, Google will have a difficult time convincing the courts that [their] AI-generated works aren’t infringements.”
Lawyers from creative companies may have a strong case if they argue that Google profits from the use of copyrighted characters. Google makes money from ads and user data, which could support a case against it if pursued by creative companies.
If you’re an author or publisher that has had your copyrighted content scraped by Google’s AI, you can find out how to report it here.
