As generative AI technologies such as GPT-4 and Midjourney rapidly become more sophisticated and their creative uses grow in popularity, the US Copyright Office has issued the directive today To show when AI-generated material can be copyrighted.
The guidelines come after the Copyright Office determined that the author could not copyright individual AI images used to illustrate a comic book, because each image was created by Midjourney — not a human artist. In making its decision, the Copyright Office committed to upholding the longstanding legal definition that authors of creative works must be human to register works. For this reason, officials emphasized that AI technologies cannot be considered authors.
This wasn’t the only case that affected the new guidelines, but it was the most recent. Wrestling with complex comic book authorship questions helped prompt the Copyright Office to launch an agency-wide initiative to continue exploring the wide range of emerging copyright issues as the AI models used to create text, art, audio, and video continue to evolve.
The guidelines provide some detail about what does not qualify for copyright when it comes to AI works created solely through claims – without modifications – which the Copyright Office likens to giving “instructions to an authorized artist.” These works lack copyright and therefore will not be registered.
When an AI technology “receives only a directive from a human and produces complex written, visual, or musical works in response, the ‘traditional elements of composition’ are determined and performed by the technology—not the human user,” the directive explains. “Based on the Bureau’s understanding of currently available generative AI technologies, users do not exercise ultimate creative control over how these systems interpret prompts and generate materials.”
However, as in the case of Midjourney, an author who arranges a generative AI into a specific sequence—such as designing a layout for a comic book—can reserve the copyright to that series of images, if the arrangement is “creative enough”. Similar reasoning applies if an author or artist modifies AI-generated material, and “the modifications meet the criteria for copyright protection.” Examples are modifying an AI image in Adobe Photoshop, or changing the AI-generated sound with guitar pedals, depending on the instructions.
It is clear, however, that the Copyright Office is only in the beginning stages of dealing with these complex cases, and the guidelines are still somewhat vague. Ultimately, officials considering whether AI-powered businesses were designed by humans or machines will reach decisions on a case-by-case basis, the guidelines say.
“The answer will depend on the circumstances, particularly how the AI tool works and how it is used to create the final work,” the directive says.
Any AI-generated content must be disclosed
Perhaps the most important aspect of the directive is the “author’s duty to disclose that AI-generated content is included in the work submitted for the recording”.
When recording works, authors must distinguish between human-authored content and content generated by artificial intelligence. If applicants are not sure how to refer to AI-generated content, the Copyright Office recommends filing a public statement that the work contains AI-generated content. This will prompt the office to proceed to assist each author in filling in the blanks on the application.
For artists who have pending applications or have already registered works that contain AI-generated content, the Copyright Office is proposing to correct the public registry by filing an additional registration. The office warned that any failure to accurately articulate the role of artificial intelligence in copyrighted works could result in a “loss of the benefits of registration”. This could leave the works open to copying, with little or no legal recourse for claims of copyright infringement.
Non-disclosure of AI-generated content is the only type of breach discussed in the guide. Critics like Alex J. Champandard, co-founder of Creative.ai — a group of hackers and artists interested in generative AI —chirp To say that current guidance puts the authors in a precarious position.
“By exposing the AI you expose yourself to breach but by not exposing the AI it is safer but it violates [the US Copyright Office]! Champandard’s tweet suggested.
Ars was unable to reach Champandard to discuss other concerns his group had about the directive.
More guidance coming in 2023
The Copyright Office knows it has a lot of work to do to clarify when AI-powered content is eligible for registration. To that end, it has planned a series of hearings in April and May to distribute public input. The office too Launch a web page Dedicated to posting updates on AI news and events, so it’s easy to keep up with the guidelines as the rules are updated.
On April 19, the hearings began with an event dedicated to literary works generated by artificial intelligence. This was followed by a session on visual arts on May 2, audiovisual works on May 17, and music and audio recordings on May 31.
Artists, creative industry stakeholders, AI developers, AI researchers, and lawyers are encouraged to do so Register to attend these sessions. But because attendance will necessarily be limited, the Copyright Office also noted in guidance published today that there will be additional opportunities for stakeholders to subsequently influence attendance at these sessions.
“The office intends to publish a notice of inquiry later this year to seek public input on additional legal and policy topics, including how the law applies to the use of copyrighted works in AI training and the resulting processing of output,” according to the directive.
It will likely be a daunting task in developing guidance, but Nora Sheland, public affairs specialist at the US Copyright Office, tells Ars that the office is “excited” to launch a broader initiative to explore artificial intelligence and will look forward to public input and feedback over the coming months.”
. “Friendly writer. Unapologetic student. Typical internetaholic. Social media ninja. Communicator. Music fanatic.”