Can You Copyright AI-Generated Content?
A great deal of news and public discussion has been fixated on the increasing popularity of generative AI — a category of artificial intelligence algorithms that can generate content (textual, visual, or audio) by “training” on vast quantities of preexisting data. Some generative AI models — including OpenAI’s ChatGPT and DALL-E models, as well as Midjourney and others — generate content in response to human-entered textual instructions, or prompts.
ChatGPT in particular has become extremely popular, attracting over one million users within a week of its November 2022 launch and more than 100 million users within two months of its launch — the fastest growing consumer technology in history.
With the proliferation of generative AI models have come attempts to register copyright in works created through use of such models. These attempts have led the U.S. Copyright Office to grapple with questions about whether and to what extent AI-generated content is protected by copyright.
In February 2023, the Copyright Office concluded that a previously issued copyright registration for a graphic novel, Zarya of the Dawn, containing images generated by the AI model Midjourney, would be limited to the human-authored text and the author’s “selection, coordination, and arrangement” of the novel’s written and visual elements, but would not extend to the AI-generated images because they lacked any human authorship — a current requirement for copyright protection.
In response to these developments, the Copyright Office recently provided its first-ever formal policy statement on the registration of AI-generated works, titled “Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence.” The purpose of the policy statement, according to the Copyright Office, is to “clarify its practices for examining and registering works that contain material generated by the use of artificial intelligence technology” and to clarify “how the Office applies copyright law’s human authorship requirement to applications to register such works.”
The policy statement explains that in evaluating AI-generated works submitted for registration, the Copyright Office asks “whether the ‘work’ is basically one of human authorship, with the computer [or other device] merely being an assisting instrument, or whether the traditional elements of authorship in the work (literary, artistic, or musical expression or elements of selection, arrangement, etc.) were actually conceived and executed not by man but by a machine.’’ Works whose “traditional elements of authorship were produced by a machine” cannot be registered because they lack human authorship. For example, “when an AI technology receives solely a prompt from a human and produces complex written, visual, or musical works in response, the ‘traditional elements of authorship’ are determined and executed by the technology — not the human user.” Thus, the statement makes clear that content created by “the generative AI technologies currently available” (including ChatGPT, DALL-E, and Midjourney) are not registrable because users of these technologies enter prompts that are more akin to “instructions to a commissioned artist” than to the exercise of “ultimate creative control.”
The statement recognizes, however, that some works containing AI-generated material “will also contain sufficient human authorship to support a copyright claim,” such as when a human selects or arranges AI-generated material in a sufficiently creative way or modifies material originally generated by AI technology. In these cases, copyright protection and registration will extend only to the “human-authored aspects” of the work.
In its guidance for copyright applicants, the statement emphasized that “applicants have a duty to disclose the inclusion of AI-generated content in a work submitted for registration and to provide a brief explanation of the human author’s contributions to the work.” This duty extends to previously filed applications and even previously issued registrations, which must be corrected to the extent they failed to disclose AI-generated material. Failure to make these required disclosures may result in the cancellation of a registration.
Ultimately, whether a particular work that includes AI-generated content may be registered for copyright is a highly fact-specific inquiry that the Copyright Office is not necessarily well suited to resolve. More concrete guidance may only come as disputes over registration of AI-generated works make their way through the U.S. courts.
And this is not the only issue on which further guidance will likely be necessary. Indeed, the policy statement acknowledges that “AI-generated works implicate other copyright issues not addressed in this statement” and that it “has launched an agency-wide initiative to delve into a wide range of these issues.” Among these issues are how copyright law applies to copyrighted works used in AI training and AI-generated content that itself includes already copyrighted works. The Copyright Office expects to seek public input on these and other issues this year.