You can’t copyright AI drivel. So don’t publish AI drivel.
A district court judge has confirmed a ruling by the US Copyright Office that AI-generated art cannot be copyrighted. That’s yet another reason you should be using AI as a tool, not a text-generator.
The district court ruling applies to art, but will likely extend to text as well
In her decision, Judge Howell wrote that copyright has never been granted to work that was “absent any guiding human hand,” adding that “human authorship is a bedrock requirement of copyright.”
The artist who was seeking copyright on his AI-generated art will, of course, appeal. And it’s not clear what the final ruling will be, but for now, AI output isn’t eligible for copyright.
While Judge Howell’s ruling applies to graphics, the same reasoning is likely to apply to text. If you publish the raw output of an AI tool, you can’t copyright it.
People working with AI tools like ChatGPT know that crafting prompts requires both effort and skill. You can work with an AI to iteratively improve its output. But if all you do with that output is to publish it verbatim, you likely won’t be able to copyright it.
The level of effort or skill isn’t an issue. Imagine, for example, that you publish a list of the 100,000 most common words on the internet. Creating such a work and publishing it is a lot of effort, but it likely doesn’t qualify for copyright either.
What this means for you as a writer
Use AI to vet your arguments.
Use AI to summarize research for easier consumption.
Use it to help find grammatical errors.
AI is a great tool for authors. But it’s not a substitute for writing. As Cory Doctorow points out, all AI-generated text is in the public domain.
AI produces hack-level, predictable, inaccurate prose. No self-respecting writer should produce such drivel.
And now, since you won’t be able to copyright it, that’s one more reason not be be lazy enough not to write things yourself.
Write like a human. You owe it to yourself. And you owe it to your readers.
Raw is the lede.
This is unenforceable — the expectation that the Copyright Office will always be able to determine what is AI generated and what is not is unreasonable.
That’s not how it works. The copyright office will grant copyright without knowing. But if someone violates that copyright and the violator says “It’s not copyrightable, it’s AI generated,” then there is no defense. The point is that AI-generated content will not have a *defensible* copyright.
I’ve been experimenting with a rough draft of poetry created by bing that I edit and then publish. I save all the history of edits from the drafts to the final product.
I’m curious if this will be defensible
If you edit it sufficiently, it should be defensible.