Sarah Silverman sues OpenAI and Meta for leading misinformation
She claims that her copyright was violated in the training of the companies’ artificial intelligence systems by ChatGPT maker OpenAI and technology giant Meta.
By analysing large datasets of human text, systems like ChatGPT learn to mimic human language.
The Meta team declined to comment. BBC questions have not yet been answered by OpenAI.
Besides Ms Silverman, two other authors are bringing the class-action lawsuit.
According to the lawsuit against OpenAI, “their copyrighted materials were ingested and used for training ChatGPT without their consent”.
Meta, which owns Facebook and Instagram, is being sued for its artificial intelligence system LLaMa. The system was initially released to a small group of researchers. It was subsequently leaked online.
LLaMa is a “foundational large language model” designed to aid AI research. Essentially, it’s a very large AI system that can be applied to many different tasks.
LLaMa was trained using a dataset compiled by another organization, according to the authors.
According to Patrick Goold, a reader in law at City University in London, both cases will likely come down to whether training a large language model constitutes fair use.
A previous lawsuit, brought by two authors against OpenAI, is already being handled by Matthew Butterick and Joseph Saveri.
“Since the launch of OpenAI’s ChatGPT system in March 2023, writers, authors, and publishers have expressed concern about its uncanny ability to create text that is similar to that found in copyrighted textual material, including thousands of books,” they write.
However, other legal experts have questioned whether OpenAI copied books.
It filed two lawsuits last year on behalf of programmers and artists who believe their rights have been violated by artificial intelligence systems.