Author Sarah Silverman and two other people filed a complaint against OpenAI and Meta for allegedly exploiting copyrighted material without authorization to train their AI systems.
Authors Richard Kadrey, Christopher Golden, and comedian Sarah Silverman have filed lawsuits alleging copyright infringement against LLaMa from Meta Platforms and ChatGPT from OpenAI. The claimants assert that Meta and OpenAI exploited their content without their consent to train their AI systems.
Many of the plaintiffs’ copyrighted publications are alleged to be present in the data set that Meta acknowledged to utilizing for training LLaMa in the court documents brought against her. This suggests that Meta unlawfully utilized the plaintiffs’ content.
Similar arguments are made in the complaint against OpenAI, where it is claimed that when ChatGPT produces summaries of the plaintiffs’ work, it shows that the training process used content that is protected by copyright. The generated summaries are said to be accurate overall, yet certain details may be wrong because information from several sources was mixed in.
The complaints assert that “shadow libraries” including Bibliotik, Library Genesis, Z-Library, and others provided Meta and OpenAI with the copyrighted data. These “shadow libraries” are websites that distribute large numbers of books via torrent protocols. Unlike open-source databases like Gutenberg that gather books with lapsed copyrights, they are regarded as unlawful sources of copyrighted content.
These shadow libraries have drawn interest from the AI training community because of the vast amount of protected content they contain. However, copyright infringement concerns are raised when copyrighted content from these sources is used without the appropriate authorization.
The authors filed the complaints on behalf of a group of copyright owners whose works were allegedly infringed upon in the United States in addition to their individual claims of copyright infringement. This expands the purview of the complaints to include additional impacted copyright holders.
The legal battles serve as a reminder of how crucial it is to uphold copyright and secure appropriate consent before using protected content for AI training. In order to defend their rights as well as the interests of other copyright owners, the plaintiffs contend that Meta and OpenAI’s conduct amount to copyright infringement.
Important issues regarding the legality and morality of using copyrighted content to train AI systems are brought up by these litigation. They underline the need of protecting intellectual property rights in the context of AI development and draw attention to the need for explicit guidelines and authorization when including copyrighted information into AI models.