If you’ve spent anytime online over the past few months, you’ve undoubtedly come across AI creations that are at once amusing and offputting—computer-generated images created from human-input prompts that look and read like something refracted through a series of funhouse mirrors. It all makes for a bit of fun for people looking to distract from the bleak online discourse we’re used to, but to those who pay attention to intellectual property, it’s a more concerning development.
In a recent op-ed in Barron’s, Shawn C. Helms and Jason D. Krieser cite the most popular of these AI bots, ChatGPT, as a potential harbinger for future copyright chaos. As with most current rules and regulations, copyright law doesn’t account for creation beyond that done by humans—a fair limitation, given that AI being able to generate creative works seemed like something out of futurist fantasy a scant few years ago. But AI-generated content is here to stay; not only that, but those creations should only improve and hew closer to actual human artistic works as the technology improves and the respective AIs learn more.
As it stands, good old-fashioned human creations seem unlikely to be affected, as current copyright law will still continue to apply. (Although to what extent companies might try to outsource that work from humans to AI is another question.) But where does the work created by Chat GPT and others stand in the current copyright landscape? Could someone type in a prompt and then take the resulting image and claim copyright on it? They’re not the creator per se, but they have more claim over that given creation than anyone else…unless you start considering the images and text that the AI is drawing from. So it’s easy to see how this can get complicated.
At the conclusion of their article, Helms and Krieser call for updated to the Copyright Act, and while that’s the logical conclusion, there’s much to be figured on long the way. Who is going to own AI-generated works, if anyone? And how can we prevent some enterprising party from trying to churn out hundreds if not thousands of images to try and stake claims? There’s much to be decided upon, and while it seems like Congress is starting to grapple with the question, it could take years before we get a handle on it.