I recently went to a conference where I spoke with a literary agent about how publishers are reacting to AI and copyright issues.
For those new to this conversation, the Copyright Office and the courts have both stated that AI-generated works cannot receive copyright protection. They’ve held this stance since AI entered the picture, and it seems unlikely they will change it anytime soon.
The Copyright Office has said that authors (their term for all creators of copyrightable works) can use AI as part of their creative process, as long as a human makes the creative decisions and the AI program is simply a tool to implement them. Practitioners refer to these as “AI-assisted” works, as opposed to “AI-generated” works. The problem is that the line between AI-assisted and AI-generated is very blurry, and we don’t yet have clarity on which side of the line certain activities fall on. For example, you can’t ask AI to write your prose; that’s a clear no-no. But what if you take the AI-generated prose and completely rewrite it in your own words? What if you feed it your words and the AI program suggests a better way to phrase things? Are these okay? At this time, we don’t have an answer.
This leaves publishers in a very difficult situation. On the one hand, many authors have at least experimented with AI by this point, so finding a manuscript that has never been touched by an AI program is becoming increasingly difficult. On the other hand, AI-generated text won’t get copyright protection, which makes the work less valuable and threatens their bottom line. And figuring out whether a text is AI-generated is problematic as well—pure AI-generated text has a certain style that industry professionals have learned to recognize, but modified text is trickier to flag. So where does that leave them?
Unfortunately, some publishers seem to be reacting with a scorched-earth policy. They will not touch any book they suspect is AI-generated or even AI-assisted. Several recent news cases show authors having their contracts canceled over suspicions that their manuscripts were AI-generated. One publishing house has even added language to its contract allowing it to terminate the contract if AI was used at any point in the creative process, including research and brainstorming. Thankfully, this does not extend to the entire industry, but it suggests a concerning trend.
So where does that leave authors who have used an AI program, or want to?
An author who wants to be traditionally published needs to tread carefully. Any use of AI programs should be documented, and they should be able to point to which parts of the manuscript could arguably be considered AI-generated. They should do research before submitting to publishers to gauge the publisher’s comfort level with AI-assisted works.
Otherwise, we need to wait and see where the industry goes from here. This is still the very early days of AI-generated content, and everyone is still figuring out what is safe and what is not. Hopefully, time will bring more clarity to the issue and allow publishers to adopt more nuanced guidelines.
If you have questions about publishers and AI or want to determine whether your use of AI programs is problematic, feel free to contact me at kaway@kawaylaw.com.
