The US Copyright Office has opened up a public consultation about all the copyright issues raised by artificial intelligence and especially generative AI. It is now inviting any interested parties to submit their opinions and positions, identifying four main topics for debate and posing a stack of more specific questions.
Its formal announcement in the Federal Register states that, to “help assess whether legislative or regulatory steps in this area are warranted”, the Copyright Office “seeks comment on these issues, including those involved in the use of copyrighted works to train AI models, the appropriate levels of transparency and disclosure with respect to the use of copyrighted works, and the legal status of AI-generated outputs”.
The American music industry is sure to have lots of opinions to share. There has, of course, been lots of debate in recent months within the music community about the challenges and opportunities posed by AI, including the AI models that generate music.
For the music industry at large, the first of the four topics identified by the Copyright Office is probably the priority. That relates to when an AI company uses existing copyright-protected works – for example songs and recordings – to train its generative AI models.
The music industry – and other copyright industries – are adamant that any such training requires consent from the copyright owner and, without that consent, the AI company is liable for copyright infringement. But not all AI companies agree.
“The Office is aware that there is disagreement about whether or when the use of copyrighted works to develop datasets for training AI models is infringing”, the Office’s announcement says.
“This notice seeks information about the collection and curation of AI datasets, how those datasets are used to train AI models, the sources of materials ingested into training, and whether permission by and/or compensation for copyright owners is or should be required when their works are included”.
“To the extent that commenters believe such permission and/or compensation is necessary”, it adds, “the Office seeks their views on what kind of remuneration system(s) might be feasible and effective. The Office also seeks information regarding the retention of records necessary to identify underlying training materials and the availability of this information to copyright owners and others”.
Another big debate within the music community – and especially among music-makers – relates to the AI models that can imitate the voice or likeness of established artists.
That is another of the four topics being considered by the Copyright Office, even though – as it notes – “these personal attributes are not generally protected by copyright law”, and instead “their copying may implicate varying state rights of publicity and unfair competition law”.
The other two topics identified by the Copyright Office relate to the copyright status of AI-generated works and who might be liable if an AI model generates content that arguably infringes copyright in one way or another.
The Copyright Office has already given quite a lot of consideration as to whether works generated by AI should enjoy copyright protection, because in the US you need to register works to get full protection under copyright law, and the Office has had to decide whether to allow AI-generated works to be registered.
Its position to date has generally been that works entirely generated by AI do not enjoy copyright protection and that position was backed by a US court earlier this month.
However, AI-assisted works – where a human creator employs AI tools as part of the creative process – probably do enjoy copyright protection. That, of course, raises further questions about how much human involvement there needs to be for a work to be AI-assisted rather than AI-generated.
“Although we believe the law is clear that copyright protection in the United States is limited to works of human authorship”, the Office’s announcement explains, “questions remain about where and how to draw the line between human creation and AI-generated content”.
“For example, are there circumstances where a human’s use of a generative AI system could involve sufficient control over the technology, such as through the selection of training materials and multiple iterations of instructions, to result in output that is human-authored?”
The final topic has not been so widely discussed in the music industry, though as the music community starts to increasingly employ AI tools as part of the music-making process, or to create visual, video or text content as part of marketing activities, it’s an important question.
If an AI tool generates a piece of content that arguably infringes copyright – either because the dataset used to train the AI model was not properly licensed or because an output is substantially similar to an existing work in the dataset and such similarity is not allowed by any licence – who is liable for the alleged copyright infringement?
Which is to say, could the person who used the AI tool also be liable? The Office notes: “If an output is found to be substantially similar to a copyrighted work that was part of the training dataset, and the use does not qualify as fair, how should liability be apportioned between the user whose instructions prompted the output and developers of the system and dataset?”
Interested parties have until 18 Oct to make written submissions.
The US Copyright Office study – part of an AI initiative it launched earlier this year – follows recent hearings in US Congress that also put the spotlight on the copyright questions posed by artificial intelligence and especially generative AI.
Those questions are also big talking points in the UK, of course, although with one difference, in that under UK law AI-generated works do arguably enjoy copyright protection.
As for the other issues, the Intellectual Property Office recently convened a working group of organisations representing copyright owners and the tech sector to develop of code of practice around AI and copyright.
That follows a previous IPO review that proposed introducing a new copyright exception into UK law that would have reduced the obligations of AI companies making use of copyright-protected works, a proposal that was strongly opposed by the music and wider copyright industries, and subsequently dropped.
The culture select committee in the UK Parliament has also just published a report that includes some discussion on copyright and AI. It welcomes the government’s back-tracking on its previous proposal of introducing a new copyright exception. However, it also criticises the fact the exception was ever proposed.
The fact it was, the MPs write, “shows a clear lack of understanding of the needs of the UK’s creative industries. All branches of government need to better understand the impact of AI, and technology more broadly, on the creative industries and be able to defend their interests consistently”.
And, more generally, “the government should support the continuance of a strong copyright regime in the UK and be clear that licences are required to use copyrighted content in AI. In line with our previous work, this Committee also believes that the government should act to ensure that creators are well rewarded in the copyright regime”.