Sarah Silverman sues OpenAI and Meta for copyright infringement
Comedian Sarah Silverman and authors Christopher Golden and Richard Kadrey are suing Meta and OpenAI over dual claims of copyright infringement. Their lawsuit accuses Meta’s LLaMA and OpenAI’s ChatGPT trained on illegally-acquired datasets containing their works. Specifically, they allegedly came from “shadow library” websites like Bibliotik, Library Genesis, and Z-library.
Generative artificial intelligence is making it harder to handle copyright infringement cases. Nowadays, AI tools can create text and other media by themselves, muddling who truly creates AI-generated content: user prompts or the machine.
Nevertheless, artists like Sarah Silverman are taking a stand to protect their works against AI. Expect such issues to become more common as AI tools improve. Learn more about their potential risks by reading this article. Later, I will elaborate on this issue and explain how you could protect your work.
Why did Sarah Silverman file a lawsuit?
— Forbes (@Forbes) July 10, 2023
The Verge shared the authors’ original lawsuit documents and explained them online. The suit claims social media company Meta and AI firm OpenAI trained their AI models on their content without permission.
ChatGPT and LLaMA reportedly trained on illegally-acquired datasets containing their content. The plaintiffs noted they are “available in bulk via torrent systems and shadow library websites.”
Sarah Silverman, Richard Kadrey, and Christopher Golden offered exhibits showing ChatGPT will summarize their books when prompted, violating their copyrights.
They cited Silverman’s “Bedwetter,” Kadrey’s “Sandman Slim,” and Golden’s book, “Ararat,” as examples. Moreover, they claimed the chatbot did not “reproduce any of the copyright management information Plaintiffs included with their published works.”
Meanwhile, their Meta lawsuit claims their books were in the social media firm’s datasets for training its LLaMA models. Also, the document elaborates why the plaintiffs believe the datasets have illicit origins.
It cited one of the company’s sources for training datasets called ThePile. Then, the complaint cites an EleutherAI paper saying it formed from “a copy of the contents of the Bibliotik private tracker.”
You may also like: Steam bans games with copyright-infringing AI content
The lawsuit claims Bibliotik and the other “shadow libraries” it mentioned are “flagrantly illegal.” Both claims say the authors “did not consent to use their copyrighted books as training material.”
Each lawsuit contains six counts of negligence, unjust enrichment, unfair competition, and copyright violations. The Verge says the authors seek statutory damages, restitution of profits, and more.
Lawyers Joseph Saveri and Matthew Butterick represent the three authors and have written on their LLMlitigation website about similar complaints. They allegedly spoke with “writers, authors, and publishers who are concerned about [ChatGPT’s] uncanny ability to generate text similar to that found in copyrighted textual materials, including thousands of books.”
Why is AI copyright complicated?
Artificial intelligence tools like ChatGPT and DALL-E have made copyright claims confusing. This problem stems from how generative AI tools work.
These programs create content based on your description. For example, you can ask ChatGPT to create a research paper about Bitcoin and other cryptocurrencies.
Consequently, it will write one with minimal grammatical and factual errors. The question is, who created the paper, you or the machine? Some argue it is the former because AI results still depend on user prompts.
You must know the right things to tell a chatbot to create something you want. On the other hand, others say the AI chatbot owns that crypto paper.
You didn’t write the paper; the program did. Let’s say the copyright “belongs” to the chatbot. That means its creator would possess copyright, correct?
You may also like: OpenAI is making a new copyright-friendly ChatGPT
Yet, the users were responsible for causing the company’s chatbot to produce content. If AI-generated content violates copyright, who should take accountability?
Most laws aren’t prepared to handle AI-generated content. For example, the United States Copyright Office registers “an original work of authorship, provided that the work was created by a human being.”
As we’ve established, AI generators make it difficult to determine whether humans are responsible for their content. Hence, countries may need to modernize their laws.
Sarah Silverman, Christopher Golden, and Richard Kadrey sued Meta and OpenAI for allegedly violating their copyrights. You may learn more about the lawsuit here.
Artists could protect their work by minting their works into non-fungible tokens. However, we still do not have a guaranteed method of maintaining AI copyrights at the time of writing.
As mentioned above, our countries must pass new laws to protect citizens’ intellectual property. More importantly, you should learn more about the latest digital tips and trends from Inquirer Tech.