AI content is on everyone’s minds. The rapid development of AI writing tools means that everything we know about AI is changing fast — and it doesn’t show signs of slowing down. But don’t get so wrapped up in the how of AI content that you forget about some of the potential consequences. Yes, AI can help you create thousands of blog posts at record speeds, but there are SEO implications, quality concerns, and massive ownership question marks to contend with.
What is copyright, anyway?
Copyright, according to the U.S. Copyright Office, “is a type of intellectual property that protects original works of authorship.” Ultimately, this means that the owner of a copyrighted work controls the reproduction, distribution, and use of their property — be that art, music, film, writing, or otherwise. Copyright law is constantly evolving, with updates to Title 17 as recent as October 2022.
Only humans can hold copyright ownership
You know the saying monkey see, monkey do? Turns out, when monkeys see humans take selfies, they want in on the action. At least, that was the case for crested macaque Naruto, who snapped a selfie that sparked copyright debate.
The macaque in question. Photo courtesy of Wikimedia Commons
Long story short: PETA sued nature photographer David John Slater in 2015 for including Naruto’s pearly white smile in a photography book and allegedly infringing on Naruto’s copyright to the photos. But according to the 9th Circuit Court of Appeals after reviewing the case, animals can’t hold copyrights.
To back it up, Section 313.2 of the Compendium of U.S. Copyright Office Practices states, “To qualify as a work of ‘authorship’ a work must be created by a human being.”
PETA lost, a new copyright precedent was sent, and now this incredible monkey selfie lives in the public domain.
What does that mean for AI-generated content?
AI is also not a human and, therefore, can’t hold a copyright.
However, the Compendium also makes a special mention of machine-created works: “Similarly, the Office will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.”
Which begs the question: how much creative input or intervention from a human author is necessary to qualify for a copyright?
The answer is still unknown.
What does OpenAI’s terms of service say about who owns generated content?
If you haven’t used OpenAI’s services for yourself yet, GPT-3 and ChatGPT are both fairly straightforward. You put a request in (“Input”), and the program spits out a response (“Output”). So, where does your ownership begin and end?
OpenAI’s terms of service states:
OpenAI hereby assigns to you all its right, title and interest in and to Output. OpenAI may use Content as necessary to provide and maintain the Services, comply with applicable law, and enforce our policies. You are responsible for Content, including for ensuring that it does not violate any applicable law or these Terms.OpenAI’s terms of service
It sounds like great news if you’re anticipating producing content using AI tools. You own both what you put in and whatever intellectual property rights that OpenAI has in what it sends out on the other side.
But there are a couple hiccups.
First, OpenAI may not have any ownership rights to transfer because the output was generated by a non-human.
Then, in the next section of their terms of service, they go on to say:
Due to the nature of machine learning, Output may not be unique across users and the Services may generate the same or similar output for OpenAI or a third party. For example, you may provide input to a model such as “What color is the sky?” and receive output such as “The sky is blue.” Other users may also ask similar questions and receive the same response. Responses that are requested by and generated for other users are not considered your Content.OpenAI’s terms of service
Which effectively means that OpenAI isn’t liable if someone else’s prompt creates a piece of copywriting that is identical to yours. And because their output is theirs and not yours, you have no authority over what someone else produces, and your ability to claim an infringement on your Output is reduced to zero. Plus, unless you personally know every user generating content using ChatGPT, you’ll have no idea if OpenAI gave a duplicate Output to someone else.
So, what happens if you end up with identical copywriting to three other sites in your niche also using ChatGPT for content?
I sat down with Verblio’s CTO and COO Greg Levow to hear his take on this constantly changing playing field, and he summed up the current status by saying, “People will be able to publish and use [AI-written] content without fear of infringing upon someone else’s intellectual property rights, but they will not be able to prevent others from using and modifying that content.”
It’s worth noting again that AI tools may provide content that replicates another source without your knowledge, so keeping humans in the process is a must for truly unique content.
What are the consequences of publishing content you don’t own?
If a human writer looks at another human writer’s work, tweaks a few words to skate past a run-of-the-mill plagiarism detector, and claims the words are their own? It’s plagiarism.
If MLAI does it… it’s the future of content creation?
As far as plagiarism is concerned, this rapidly developing AI situation is a gray space where claiming ownership is challenging because the parameters the models were trained on include web pages from across the Internet. Without consent.
(This isn’t just for AI-generated writing, BTW. Copyright infringement cases are cropping up against AI-generated art, most notably Getty Images suing Stable Diffusion for unlawfully training their models on watermarked images, and they aren’t the only ones.)
Because OpenAI’s machine learning models were trained on published web pages without the express permission of the owners of those web pages, it’s arguable that AI is inherently plagiarized. Unlike Stable Diffusion, an open source company, OpenAI’s training sets aren’t publicly available. Right now, it’s impossible to know whose content has been soaked up by the web of neural networks used by OpenAI’s product line.
Content spinners and blatant cheaters are plagiarizing content from the internet all the time, but AI will likely make this far worse—and with less recourse. Verblio has a strict no plagiarism policy on our marketplace, and responding to the influx of AI-generated content is one of our highest priorities.
How does Verblio ensure actual humans are creating our human-written content?
Human-written content still has its place in the world. If you’re even considering posting thought leadership, don’t let an AI do it for you.
At Verblio, we’ve tested gobs of AI detection tools. Finding the best one meant looking at several factors:
- Accuracy detecting AI content
- Integration with our systems
- Ability to scale
Originality.ai is our tool du jour, but as tools continue developing, we’ll continue testing them.
To protect our customers, we’ve already employed an additional level of security by filtering every piece of content submitted on our marketplace through this program. This is being closely monitored by our internal teams so that we can quickly and easily address any infractions and provide customers with original (human-produced) works only.
How is Verblio’s Human-Crafted AI Content different from other AI content?
ICYMI: Human-Crafted AI content is our response to the AI content revolution. It puts the power in the hands of expert writers and uses AI as a tool to impact speed and cost.
HCAI is different than plugging a few prompts into ChatGPT and pasting the outputs straight onto your blog.
At every step of the process, humans intervene to curate, edit, and fact check the content. Much like throwing clay with a potting wheel, integrating AI into our workflow is done for the sake of expediting the creative process—not replacing it.
How do we address copyright at Verblio?
If you’re a Verblio customer, you’re probably wondering how we’re moving forward with regard to our human-produced content, our HCAI product, and the onslaught of AI tools.
We do not allow purely AI content to be submitted through our marketplace. Pure AI content doesn’t live up to our quality standards. Our human-only content will stay human-only.
What comes next for AI content?
Honestly? The AI copyright landscape is messy right now. It’s going to stay messy for a while. As we (both at Verblio and in society as a whole) learn how reproductions with regard to AI will be handled, the future isn’t crystal clear.
For now, our takeaways are this:
- You don’t hold the copyright to AI content. Neither does the AI.
- Pure AI content can be replicated without recourse.
- Human-Crafted AI keeps the control in the hands of humans.
We’re keeping our finger on the pulse of AI content developments. Whatever happens, we’re confident AI will revolutionize the way we create, distribute, and think about content.