However, legal experts argue that current laws are ill-equipped to address the use of copyrighted data for AI training and that this issue needs to be dealt with at a policy level. They warn that the continued uncompensated use of creative works for AI training could discourage creators from making their work available online, potentially starving AI of training data. They suggest that legislators need to articulate a new vision for balancing rights and responsibilities, similar to the response to the development of the internet.
Key takeaways:
- Microsoft AI's CEO, Mustafa Suleyman, stated that machine-learning companies can use most online content to train neural networks as it's essentially 'freeware.'
- Several organizations, including the Center for Investigative Reporting and the New York Times, have sued OpenAI and Microsoft for allegedly using their content without permission or offering compensation.
- Suleyman acknowledged the controversy and attempted to differentiate between content people put online and content backed by corporate copyright holders, stating that the latter is a gray area that will likely be resolved in court.
- Frank Pasquale and Haochen Sun, law professors at Cornell Tech and The University of Hong Kong respectively, argue that the legal uncertainty surrounding the use of copyrighted data to train AI needs to be addressed at a policy level, as current laws are ill-suited to answer the emerging questions.