I thought, “How cool would a robot being arrested for art theft be here?” But then I realized I can’t draw that well, don’t have the time to get better, and asking an AI to draw it would be unethical due to data sourcing concerns.. so here’s a robot emoji!
It’s hard enough to get politicians to acknowledge the science on serious issues like people’s healthcare, bodily autonomy, or climate change. Getting the luddites in government to understand the existential threat of AI was never going to be an easy task. AI companies are gobbling up the work of others, without permission or compensation, to recombine, repackage, and resell. It’s like robbing an artist’s laptop and selling its contents. Creators’ works, their livelihoods, their voice, is being stolen, repackaged, and resold to uncreative types to mash together with GenAI tools. If you think it’ll stop at creative works, think again. Already we’re seeing it take over the jobs of journalists, translators, graphic designers, voice actors, artists, writers. Soon, however, we’ll see it taking on more administrative tasks, like scheduling, and cross-team communications within companies, planning, and even coding. When Microsoft bought GitHub so many years ago, many of my fellow software engineer friends and I wanted to start pulling our code off GitHub because we knew what Microsoft wanted with the largest database of code on the internet: train AI to replace software engineers.
We uploaded our work to the cloud, then companies bought the cloud to steal our work and recreate it.
Of course, there will still be room for all of these jobs. AI will never be able to replace the agility and creativity of the human mind. All it can really do is rehash our work. But it’s doing so from copyrighted works without permission or compensation. With a new law proposed by Adam Schiff, a Democrat in the House from California, that could all change. His bill looks to force AI companies to disclose when they’re using copyrighted material. That could help creators pursue lawsuits against companies, like OpenAI, who now hide their sources. OpenAI didn’t always hide the sources for their datasets. Once they revealed that the largest source of copyrighted information they pulled from was The New York Times. They did so without permission or compensation. When NY Times saw ChatGPT could spit out text from articles verbatim, they had all they needed for a lawsuit. While other creators have failed in their legal pursuits, lacking this proof, NY Times could pursue a lawsuit that they could actually win. One bill may ensure that all creators have that chance.
Proposed Law
Schiff’s bill, the Generative AI Copyright Disclosure Act, sets out to “balance the immense potential of AI with the crucial need for ethical guidelines and protections,” according to its author. It’s a succinct bill that requires companies report all sources that contain copyrighted material to the Copyrights Register 30 days before the models are made public. They’ll also have to provide a link to the model’s location if it’s public. The law would be backdated, if passed, giving companies 180 days after the law is enacted to go into effect. That’s more than enough time to compile a list of sources that may contain copyrighted data. If AI companies actually believe their for-profit use of copyrighted materials without the owner’s permission or compensation is indeed legal and falls under fair use, they should have no problem complying with such a law.
“It champions innovation while safeguarding the rights and contributions of creators, ensuring they are aware when their work contributes to AI training datasets. This is about respecting creativity in the age of AI and marrying technological progress with fairness.”
– Rep Adam Schiff (D, CA)
This law wouldn’t necessarily make it illegal to use copyrighted material when making AI, but it could be the first step towards getting AI companies to behave. OpenAI previously disclosed their sources, but stopped by the time they were working on ChatGPT-4. Previous versions showed the third largest source of data they used came from The NY Times, without the publication’s permission. After NY Times found ChatGPT could output sections of their articles, verbatim, they sued. By forcing AI companies to track the copyrighted materials they use, and report it, people who had their copyrighted materials gobbled up will be able to see if their content was used without permission. If they decide to pursue a lawsuit, they’ll have proof their content is in the dataset, straight from the company that used it.
Failure to share copyright information with the Copyright Register will net fines of at least $5,000. This could be quite minimal for large companies, but depending on the size of the models, perhaps that fine could increase substantially. It could be just enough of a problem to convince companies to pay for their data and get permission to use it in advance. While copyright law hasn’t cracked down on AI yet, this could be the first step towards codifying ethical uses of people’s work in AI datasets.
Supporters
In the press release on Rep. Schiff’s page, there’s a list of supporters from across creative industries. It’s substantial, and, even to people outside of the industry, you’ll see some familiar names. I’m going to list them here, just to give you an idea as to how important legislation like this is to creative groups.
- Recording Industry Association of America (RIAA)
- Copyright Clearance Center
- Directors Guild of America
- Authors Guild
- National Association of Voice Actors
- Concept Art Association
- Professional Photographers of America
- Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA)
- Writers Guild of America West
- Writers Guild of America East
- American Society of Composers
- Authors and Publishers
- American Society for Collective Rights Licensing
- International Alliance of Theatrical Stage Employees
- Society of Composers and Lyricists
- National Music Publishers Association
- Recording Academy
- Nashville Songwriters Association International
- Songwriters of North America
- Black Music Action Coalition
- Music Artist Coalition
- Human Artistry Campaign
- American Association of Independent Music
You can expect this list to grow, and even include notable individuals. Basically anyone who is creative, anyone who relies on a base of knowledge, expertise, and talent, accrued over years of hard work, practice, and patience, will want to protect their work against AI companies looking to use it for their own profit.
Opposition
OpenAI hasn’t commented yet, and has not responded to a request for comment from Gizmodo as of this writing. They’re one of the largest creators of AI software, and used copious amounts of copyrighted data in the generation of their tools, much of which came without permission or compensation. AI companies and lobbyists representing them are likely already looking to prevent such laws, but the public will likely support them. After all, most people are humans, not AI companies using the labor of humans for free.
A Path Forward for AI and Humans, Finally
AI is going to be a huge and beneficial part of all of our lives eventually. It won’t just be a tool for the rich to acquire and sell the work of laborers without compensation, it’ll be a tool we use to enhance the knowledge we have on hand, improve our creative vision, and enrich our lives. There’s a fantastic future ahead for AI… but only if we nip these problems in the bud immediately. We cannot allow the rich companies, backed by multi-billion dollar efforts, steal the labor of others to turn around and sell back to us. We cannot allow it to endanger jobs, especially those of creative people that push forward the culture and very fabric of humanity that unites all of us. For that, we’ll need to put it in print: “AI can’t be used to sanitize IP theft.” Until that point, we can at least ensure that, once we definitively say that AI companies must compensate creators, we will have an authority that can ensure creators have been compensated and gave permission to use their work willingly. If we don’t ensure AI works for everyone, it’ll just serve the rich, widening the already shocking gap between the labor class and those who simply make money from other people’s work.
The AI apocalypse won’t be Skynet robots destroying humanity, it’ll be corporations erasing the value of human labor in the name of profit. Artists will find new and creative ways to poison datasets and ruin the prospects of AI if we can’t all get along. The only way to make AI is to make sure it works for everyone, not just the rich, but the people who actually create value. That will require a dramatic shift in AI’s current trajectory. Perhaps this bill will be the start of that shift… if we can get it signed into law.
Sources:
- Emilia David, The Verge
- Matt Novak, Gizmodo
- Adam Schiff’s Website