As the incredible potential of generative artificial intelligence (AI) unfolds, a crucial question arises: Is it possible to develop comprehensive large-language models while still respecting the rights of content creators?
Last week, these three plaintiffs – Silverman, Christopher Golden, and Richard Kadrey – filed lawsuits in a federal court in San Francisco, alleging that both companies illegally incorporated their works into the training of their models. They claim that OpenAI and Meta utilized extensive digital-book collections, known as “shadow libraries,” which contain copyrighted material without proper licensing.
Of particular concern is the fact that ChatGPT can generate detailed summaries of books upon request. For instance, when prompted to summarize Silverman’s book “The Bedwetter,” it provides a thorough summary. The lawsuit argues that such a summary would be impossible without the complete text being present in the training model.
This legal action underscores the need for ethical considerations when harnessing the capabilities of AI models while ensuring respect for the intellectual property rights of content creators. The outcome of these lawsuits could potentially shape how AI models are developed and trained in the future.
Lawsuit Filed Against Meta Over Unlicensed Copyrighted Material
A recent lawsuit has been filed against Meta (formerly known as Facebook) by lawyers Joseph Saveri and Matthew Butterick. This legal action follows several other similar lawsuits filed by the attorneys, all of which focus on the use of unlicensed copyrighted material in training sets for large language models.
Neither Meta nor OpenAI have provided any official comments on the ongoing lawsuits.
It is worth noting that most creators of large-language models do not disclose detailed information about the data used to train their models. However, there is an exception in the case of Adobe. Adobe has stated that its image-creation model exclusively uses images that are licensed to the company through their stock-photography business or those that are no longer under copyright protection. Furthermore, Adobe is actively working on a method for creators whose works are included in Adobe Stock to opt out of their inclusion in training models. This development is accompanied by a compensation plan for artists, ensuring they receive proper remuneration when the software generates new work based on their creations.
