The Wikimedia Foundation, the organization supporting Wikipedia, has entered the ongoing legal dispute between The New York Times and both OpenAI and Microsoft, submitting an amicus curiae brief urging the adoption of a precise and balanced approach.
The core case revolves around whether the use of content published by the newspaper for purposes including training artificial intelligence models constitutes a copyright infringement, or falls under the concept of fair use. While the newspaper demands that companies be prevented from freely using its content, technology companies argue that training on public data is essential for technological development.
For its part, the Wikimedia Foundation proposed a third path situated between the two positions. The brief warned that imposing excessive restrictions on fair use could hinder technological innovation, while recognizing absolute fair use might lead to a closed web where knowledge is weakened due to a lack of available secondary sources.
Proposed Fair Use Applications
The Foundation defined its alternative position by applying the principle of fair use to two specific cases for artificial intelligence companies, which are:
- Certain scenarios for training large language models that produce new and different technologies from the original sources.
- Retrieval-Augmented Generation (RAG) technology, which improves model outputs by referencing reliable knowledge bases outside the training data, provided that previous articles are not reproduced without attribution and source crediting.
Stephen LaPorte, General Counsel at the Wikimedia Foundation, pointed out that issuing a ruling that does not apply fair use principles with utmost care could threaten the sustainability of free knowledge and stifle the development of open-source technologies, negatively impacting everyone's access to information.
What Does This Intervention Represent?
This step comes at a time when attention is turning toward the issuance of a future ruling in this case, which is considered one of more than ten similar lawsuits being considered by the courts. It is expected to set an important legal precedent shaping the future relationship between artificial intelligence developers and digital content producers, especially since Wikipedia itself is considered one of the most important and highest-quality databases used in training intelligent models.