By: Andrew J. Costa, Esq.
The Copyright Office’s ongoing study of artificial intelligence and its impact on copyright law and policy continues with the Office’s issuance of a wide-ranging Notice of Inquiry published on August 30, 2023 (the “Notice”). In the Notice, the Copyright Office attempts to build upon its public listening sessions from earlier this year and seeks answers to some of the most pressing AI topics that have transfixed both the legal community and the public alike. While the Notice itself, of course, provides no answers on these topics, it does demonstrate a welcome shift in the Copyright Office’s focus from one that has largely sought a general understanding of AI, towards one that appreciates the nuances of the more controversial questions; and the Office seems ready to tackle them head on.
In this post, we provide a brief overview of the Notice, highlight some questions that particularly interest us, and provide some additional resources for further reading on AI generally.
Overview of the Notice
The main purpose of the Copyright Office’s Notice of Inquiry is to assess whether regulatory or legislative action is needed to address the impact and effect of AI. Indeed, this grand question was one of the key topics discussed by lawmakers and tech leaders at the recent tech summit on Capitol Hill hosted by Senate Majority Leader Chuck Schumer (D-NY). That summit included tech innovators like Bill Gates, Elon Musk and Mark Zuckerberg, among others. While those leaders agreed that regulation is necessary, they could not agree on what that regulation would actually look like – thus making this inquiry even more timely.
The notice itself is divided into four distinct topics, each of which has raised controversy within the legal, AI and tech communities: (1) the use of copyrighted works to train AI models, (2) the copyrightability of AI-generated outputs, (3) potential liability for infringing works generated by AI, and (4) the treatment of AI-generated outputs that mimic real-life people (such as deep-fakes or software that imitates an artist’s voice or style). Helpfully, the Notice provides a general overview of each topic, before presenting the Office’s specific questions. Interestingly, the Notice also includes a glossary of various AI terms, such as “generative ai”, “training datasets”, and “weights”, reflecting the ongoing difficulty of discussing AI without consensus on definitions.
Lastly, the Notice also subtly attempts to address criticism of the Copyright Office’s recent AI decisions – i.e., Zarya of the Dawn and A Recent Entrance to Paradise – by restating some of its principals of authorship. The Notice recites the Copyright Office’s history of addressing authorship concerns, even going as far back as 1965 and the computer revolution. Quoting from that era, the Copyright Office explains (as it did then) that “[t]he crucial question appears to be whether the ‘work’ is basically one of human authorship, with the computer merely being an assisting instrument, or whether 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.” In doing so, the Office reiterates its enduring perspective that, at its heart, authorship (and hence copyright) is a distinctly human endeavor.
(You can read about our take on the Zarya of the Dawn decision here.)
The following are some of the Copyright Office’s key questions that particularly interest us. Each is identified by topic category and includes the corresponding question number from the Notice.
Use of Copyrighted Works in Training:
- [7.3] “Is it possible for an AI model to ‘unlearn’ inferences it gained from training on a particular piece of training material?”
- [7.4] “Absent access to the underlying dataset, is it possible to identify whether an AI model was trained on a particular piece of training material?”
- [8.1] “In light of the Supreme Court’s recent decisions in Google v. Oracle America and Andy Warhol v. Goldsmith, how should the ‘purpose and character’ of the use of copyrighted works to train an AI model be evaluated? What is the relevant use to be analyzed?”
- [8.3] “The use of copyrighted materials in a training dataset or to train generative AI model may be done for noncommercial or research purposes. How should the fair use analysis apply if AI models or datasets are later adopted for use of a commercial nature? Does it make a difference if funding for these noncommercial pr research uses is provided by for profit developers of AI systems?”
- [10.3] “Should Congress consider establishing a compulsory licensing regime?”
-  “What obligations, if any, should there be to notify copyright owners that their works have been used to train an AI model?”
Generative AI Outputs – Infringement
-  “Is the substantial similarity test adequate to address claims of infringement based on outputs from a generative AI system, or is some other standard appropriate or necessary?”
-  “How can copyright owners prove the element of copying (such as by demonstrating access to a copyrighted work) if the developer of the AI model does not maintain or make available records of what training material it used?”
-  “What legal rights, if any, currently apply to AI-generated material that features the name, likeness, including vocal likeness, of a particular person?”
-  “Should Congress establish a new federal right, similar to state law rights of publicity, that would apply to AI-generated material? If so, should it preempt state laws or set a ceiling or floor for state law protections? What should the contours of such right be?”
-  “Are there or should there be protections against an AI system generating outputs that imitate the artistic style of a human creator . . . [w]ho should be eligible for such protection? What form should it take?”
We are reminded of the Latin saying, prudens quaestio dimidium scientiae, i.e., “half of science is asking the right questions”. Now that the right questions are being asked, we are hopeful that we will get to the right answers.
The Copyright Office is accepting comments until 11:59 PM EST on Monday, October 30, 2023, and commentators may reply to the comments of others until 11:59 PM EST on Wednesday, November 29, 2023.
AI Registration Guidance (U.S. Copyright Office), March 2023.
An Introduction to Artificial Intelligence for Federal Judges by James E. Baker, Laurie N. Hobart and Matthew Mittelsteadt (Federal Judicial Center), February 2023.
Artificial Intelligence: Background, Selected Issues, and Policy Considerations by Laurie Harris (Congressional Research Service), May 2021.