Meta and OpenAI Accused of Using LibGen’s Pirated Books to Train AI Models

February 28, 2023

A recent investigation into the training data of Meta's LLaMA 3 and OpenAI's models has uncovered allegations that both companies used a pirated book repository, LibGen, due to licensing concerns. The dataset, which contains over 7.5 million pirated books, was accessed with apparent approval from Mark Zuckerberg. As a result, employees took steps to obscure the dataset’s origins, raising questions about the ethics of using such data in AI model training. This incident highlights the need for responsible AI practices and more stringent controls on accessing pirated materials.

The use of LibGen's pirated books in Meta's LLaMA 3 and OpenAI's models raises profound questions about corporate accountability and the ethics of AI model training. As the tech industry grapples with the consequences of this incident, it is essential to hold companies like Meta and OpenAI responsible for their actions. The fact that Mark Zuckerberg's approval led to the acquisition of pirated materials serves as a stark reminder of the need for stricter licensing controls and more transparent data sourcing practices in AI development. By acknowledging their role in this scandal, we can work towards creating a more responsible and sustainable future for AI research – space to learn more about this issue is available at JOIN US.

Matched TAIM controls

Suggested mapping from embedding similarity (not a formal assessment). Browse all TAIM controls

Alleged deployer
OpenAI, Meta
Alleged developer
OpenAI, Meta
Alleged harmed parties
Writers, publishers, Journalists, Authors, Academic researchers

AI governance case studies

For forensic AI governance failure analysis (TAIMScore™ case studies), browse Human Signal’s Failure Files™.

Source

Data from the AI Incident Database (AIID). Cite this incident: https://incidentdatabase.ai/cite/997

Data source

Incident data is from the AI Incident Database (AIID).

When citing the database as a whole, please use:

McGregor, S. (2021) Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database. In Proceedings of the Thirty-Third Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-21). Virtual Conference.

Pre-print on arXiv · Database snapshots & citation guide

We use weekly snapshots of the AIID for stable reference. For the official suggested citation of a specific incident, use the “Cite this incident” link on each incident page.