Writers Furious After Learning Their Copyrighted Work Was Used to Train Meta’s AI
Australian Authors React to Meta’s Use of Their Work for AI Training
Several prominent authors from Australia are expressing their outrage following revelations that Meta, the parent company of Facebook, has used their literary works without consent to develop its artificial intelligence (AI) platform. This situation has sparked significant debate about copyright, fair use, and the ethical implications of using creative works to train AI systems.
The Discovery of Unauthorised Use
Background of the Issue
In January, court documents indicated that Meta had utilized Library Genesis (LibGen), a well-known website featuring pirated books and academic papers, to train its generative AI language model. This revelation came to light when US magazine The Atlantic published a searchable tool for LibGen on March 20, allowing the public to explore the extensive database of pirated materials, which included notable works by Australian authors such as Charlotte Wood, Tim Winton, and Helen Garner.
Authors’ Emotional Responses
Many authors reacted strongly to the news. Sophie Cunningham, who has several works available on LibGen, including her 2022 novel This Devastating Fever, expressed her anger, stating, “It’s one thing to be underpaid. It’s another to find that your work is being used by a company you don’t trust.” She is contemplating legal action against Meta and has requested that her publishers send cease and desist letters on her behalf.
Bestselling author Hannah Kent shared her feelings of devastation upon discovering that all three of her novels were included in the LibGen database. “It felt like my body of work has been plundered,” she lamented.
Reactions from Literary Advocates
Lucy Hayward, CEO of the Australian Society of Authors (ASA), articulated her concerns over tech companies exploiting creators without compensation. She highlighted the inconsistency in how these companies operate, earning substantial revenues while authors receive negligible financial benefits from their own creations.
Licensing Complexities
Meta’s use of copyrighted material isn’t without precedent. The company previously faced scrutiny for using databases of pirated content to train AI, leading to ongoing legal challenges. In a copyright lawsuit, a disclosure revealed that Meta considered negotiating licenses with authors and publishers but deemed the process “incredibly slow” and “unreasonably expensive.” Documents suggested a preference for exploiting available data rather than obtaining it through legitimate means.
Fair Use and Legal Implications
The Fair Use Defense
Currently, Meta is embroiled in multiple lawsuits concerning copyright infringement, including the case of Kadrey v. Meta, where it is relying on the fair use defense, which allows for limited use of copyrighted materials without permission. However, experts like Toby Walsh, a leading AI researcher, argue that this defense is questionable, particularly when Meta did not attempt to obtain the content legitimately.
Walsh considers the situation a significant ethical transgression, suggesting that the actions of Meta and similar companies represent a “greatest heist in human history.” He argues that big tech firms ignore copyright laws to create profitable AI models — and that the intellectual labor of authors goes unrecognized and uncompensated.
Authors’ Call for Action
Both Sophie Cunningham and Hannah Kent share a vision for more stringent regulations governing the use of creative works in AI training. They advocate for explicit consent and appropriate compensation for authors whose works are used. Kent articulated her desire for the option to approve or decline the use of her material, emphasizing the importance of agency for creators.
Future Considerations
Anticipated Changes in Copyright Law
As legal challenges surrounding AI and copyright unfold, there may be significant changes to how AI can be trained in the future. The ongoing discussions encourage a reassessment of copyright laws that can effectively address the nuances of AI technology and protect authors’ rights.
The ASA is involved in collaborative efforts with the government to address these issues, ensuring that the interests of creators are safeguarded. Walsh likens the current era to a pivotal moment in the music industry when streaming services transitioned to a model that compensates artists for their work.
Regulating AI Development
Congressional actions and industry responses in the coming years will likely shape the landscape of AI development and copyright laws. Understanding the ethical implications and the rights of authors is critical to ensuring that creators are valued in a rapidly evolving digital landscape. Authors and advocates continue to push for better regulation, emphasizing the need for system transparency and respect for intellectual property rights.
