Last week, I gave a keynote talk at the International Association of Privacy Professionals (IAPP) conference on AI governance. Yes, there is now a field called AI governance, which refers to frameworks, policies, and best practices to ensure AI technologies are used in a way that maximizes benefits and minimizes risk. IAPP membership consists of privacy officers, lawyers, and others who help ensure businesses and institutions follow global laws and regulations regarding privacy. This is a profession that extends to all types of industries and nonprofits—health care, financial services, retailers, you name it.
The conference began just days after the Biden Administration issued an executive order on AI, which was referenced in every discussion I heard. The executive order touches on the myriad ways that society is and will be affected by AI and the protections that people deserve in the emerging AI era. While copyright and intellectual property are mentioned in the order, these issues constitute a fraction of the challenges ahead. But here’s the most relevant part for authors:
- Within 270 days of the date of this order or 180 days after the United States Copyright Office of the Library of Congress publishes its forthcoming AI study that will address copyright issues raised by AI, whichever comes later, consult with the Director of the United States Copyright Office and issue recommendations to the President on potential executive actions relating to copyright and AI. The recommendations shall address any copyright and related issues discussed in the United States Copyright Office’s study, including the scope of protection for works produced using AI and the treatment of copyrighted works in AI training.
Elsewhere, the abbreviated fact sheet noted a key objective:
- Protect Americans from AI-enabled fraud and deception by establishing standards and best practices for detecting AI-generated content and authenticating official content. The Department of Commerce will develop guidance for content authentication and watermarking to clearly label AI-generated content. Federal agencies will use these tools to make it easy for Americans to know that the communications they receive from their government are authentic—and set an example for the private sector and governments around the world.
I heard executives from Microsoft and IBM specifically touch on this during the conference: They are developing such content authentication and provenance tools, although it wasn’t clear how soon they’ll be available. And of course many are skeptical that 100 percent accuracy is possible. But the consensus seems to be that an attempt must be made anyway. (A bipartisan bill, the AI Labeling Act of 2023, also calls for transparency.)
On a publishing listserv that I belong to, I was alerted to an early watermark standard, known as C2PA, for tracing media origin. Most people I’ve talked to believe that we need some kind of verifiable, tamper-proof attribution for creators and rights holders, and not just for AI-related reasons. On that same listserv, publishing consultant Bill Kasdorf posted, “The Anti-Counterfeiting Task Force [as part of W3C] is working on the problem of books sold by parties representing themselves as the publisher and the books as the real books, when neither of those things is true. Those books often rise to the top of recommendations because they’re sold at a somewhat cheaper price and a shocking number of people buy them.” Just as a shocking number of people buy AI-generated books not knowing what they are.
This raises the challenging question of how to draw the line between AI-generated and AI-assisted content. Under today’s law, AI-generated content isn’t protected under copyright, and this issue was discussed at a conference panel I moderated called “No AI without IP.” Four lawyers tackled the critical questions still unanswered about AI: Are generative AI models infringing copyright when they ingest copyrighted works without permission? (And what are the risks if you decide to use them?) Are the outputs of generative AI models infringing copyright? Can the outputs of generative AI models be protected under copyright? If so, under what conditions? And who owns that copyright?
The Copyright Office posted a notice of inquiry on the matter, and they were so inundated with comments, they had to extend the deadline as a result. You can review comments here; there are nearly 10,000, but you can search by keyword and find comments from organizations like the Authors Guild as well as Big Tech firms.
I asked an IP lawyer at the conference if he considered generative AI too risky for businesses or individuals to use, given the unanswered questions about infringement. His answer indicated that most people are proceeding under the assumption there will not be retroactive punishment but possibly future restrictions. Companies like OpenAI, Microsoft, and Adobe are already telling their (paying) customers they will be protected from any infringement claims arising from use of their AI tools. So if you use Adobe’s Photoshop to generate some book illustrations, and it’s later found their AI tools infringe copyright, Adobe has to fight the resulting legal battle, not you. Of course OpenAI, Microsoft, and Adobe are the kinds of companies with the size and power to help nudge laws in a direction that’s favorable to them so they won’t be fighting any legal battles over this.
In the UK, some of the biggest publishers have issued guidelines on how staff should use AI. This is a best practice I heard outlined at the IAPP conference: Companies can’t just stay silent on the issue or turn a blind eye. The workforce needs to be told what’s permissible and what’s not, based on how much risk the company feels comfortable taking on. (One company rep said they told employees it’s okay to use generative AI for work, but to use their personal email addresses. Is this really an ethical or legit strategy? No idea!) Hachette UK has told its employees it’s okay to use AI for operational uses, like copywriting and metadata, but not for creative uses that would potentially replace human authors, designers, illustrators, or translators.
Bottom line: In a nutshell, no one at the conference had firm answers. It was mostly about how to proceed—how industry leaders are proceeding—while managing the risk. That doesn’t mean there aren’t strong opinions and positions about what direction things will go in. Just last week, a judge dismissed most of a copyright lawsuit by artists against generative AI companies, concluding their accusations are “defective in numerous respects.” (Learn more.) Each day at the conference, I passed by an active exhibition area full of tech companies serving the field known as AI governance. At first I was shocked—it’s only been a year since ChatGPT came into our lives, and there are already AI governance technologies? But then I thought, Oh, it’s been a full year. And of course AI in some sense has been around for a while, only called something else, like machine learning.
Jane Friedman has spent her entire career working in the publishing industry, with a focus on business reporting and author education. Established in 2015, her newsletter The Bottom Line provides nuanced market intelligence to thousands of authors and industry professionals; in 2023, she was named Publishing Commentator of the Year by Digital Book World.
Jane’s expertise regularly features in major media outlets such as The New York Times, The Atlantic, NPR, The Today Show, Wired, The Guardian, Fox News, and BBC. Her book, The Business of Being a Writer, Second Edition (The University of Chicago Press), is used as a classroom text by many writing and publishing degree programs. She reaches thousands through speaking engagements and workshops at diverse venues worldwide, including NYU’s Advanced Publishing Institute, Frankfurt Book Fair, and numerous MFA programs.
