The US Copyright Office is inching closer to creating new rules and regulations around generative AI and how the technology uses the work of authors and other creators.
In a notice filed Wednesday, the US Copyright Office said it is opening up a period of public comment, now through October 18, to further "inform" its ongoing study of artificial intelligence tools like Midjourney, OpenAI's ChatGPT, and Google Bard.
The Office is looking at possible regulatory action or new federal rules due to "widespread public debate about what these systems may mean for the future of creative industries."
The Office wants input on whether new rules or regulation for generative AI are needed, along with the issues surrounding 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."
Authors, visual artists and even source code developers are already suing the likes of OpenAI, Microsoft and Meta because their original work has been used without their consent to train something that may ultimately compete against them.
In the government rule-making process, a public comment period typically happens before a final rule is proposed and adopted. The Office earlier this year held a series of "listening sessions" with stakeholders, including representatives of Microsoft (a major backer of OpenAI), VC firm Andreessen Horowitz and The Authors Guild.
Generative AI tools including ChatGPT, Bard and Meta's Llama 2, were developed using massive amounts of information and data scraped and saved by automated web crawlers, which suck up everything they can online, including millions of works under copyright. The major tech companies behind these generative AI tools use the crawled data to train their models without paying the creators who produced the original content. These the models need high-quality, human-generated content to perform well.
More online businesses are slowly becoming aware of the degree to which the web is being scraped for the benefit of generative AI. The likes of Amazon, Airbnb and Tumblr, along with news outlets like The New York Times and CNN, are increasingly blocking web crawlers through an old and relatively feeble tool known as robots.txt. At the moment, there is almost no way to avoid content being crawled and used to create an LLM and copyright law does not currently address the issue.
The US Copyright Office noted that it has already ruled on a some AI-content issues in the last year. In one case, it refused to register a copyright for an AI generated artwork. In another, a book by a human author that included AI generated imagery, the Office only granted copyright for the text by the author.
Now, the Office is looking to go further. It wants to resolve several issues it has identified with regard to copyright and AI, including training data, whether AI generated material is eligible for copyright, liability for infringement by AI systems and how to treat AI outputs that "imitate the identity or style of human artists."
That last issue appears to be new for the Office. Although this has become another widely discussed topic due to ongoing strikes by Hollywood writers and actors, who are fighting, in part, against a desire by studios to use generative AI in place of human writers and performers.
"In both our listening sessions and other outreach, the Office heard from artists and performers concerned about generative AI systems' ability to mimic their voices, likenesses, or styles," the Office wrote in its notice. "Although these personal attributes are not generally protected by copyright law, their copying may implicate varying state rights of publicity and unfair competition law, as well as have relevance to various international treaty obligations."
