AI and Copyright: A Legal Perspective
The recent US court ruling on AI training and copyright has sparked significant discussion in legal circles. The case involved Anthropic, an AI company that trained its model, Claude, using copyrighted works. The court recognized this training as "fair use" because the original content was transformed in the process.
Senior District Judge William Alsup stated, “The training use was a fair use,” highlighting the transformative nature of the technology involved. This decision marks a critical interpretation of copyright law in the context of generative AI.
Where the Line Was Crossed
However, the court also found that Anthropic’s method of acquiring data crossed legal boundaries. The company downloaded millions of copyrighted books from pirate sites such as Library Genesis (LibGen) and the Pirate Library Mirror without paying for them. This act was deemed unlawful.
The judge noted that purchasing print copies after illegally downloading digital versions does not absolve the company from liability. Each infringement could entail penalties up to $150,000 under US copyright law, potentially costing the AI company billions.
Comparing US and Australian Copyright Laws
Unlike the US, Australian copyright law does not have a “fair use” doctrine. Instead, it follows “fair dealing,” which offers limited exemptions such as for news reporting. This distinction makes the Australian legal framework more restrictive regarding AI training on copyrighted material.
Josephine Johnston, CEO of the Copyright Agency, points out that Australia benefits from existing statutory licenses that facilitate AI development in education and government sectors. These licenses ensure fair payment to rights holders while supporting large-scale legal use of copyrighted content.
Licensing as a Path Forward
Copyright Agency recently extended its business license to cover the use of Australian news content for AI tools, including tasks like summarization and analysis. This development aims to balance the needs of AI developers with those of content creators.
Johnston emphasizes that the US fair use approach has led to uncertainty and legal challenges, as seen in the Anthropic case and others. These situations point to the risk of prolonged litigation and an unstable environment for developers.
In contrast, Australian law offers more certainty through its licensing framework, which encourages innovation while ensuring creators receive proper compensation. Johnston stresses the importance of maintaining this system rather than weakening it to favor multinational tech companies.
Implications for Legal Professionals
Legal experts advising AI companies should carefully consider the source of training data. Indiscriminate scraping of copyrighted material, especially from pirated sources, exposes companies to significant legal risk. Licensing agreements provide a clearer, safer path for AI development.
As AI continues to evolve, balancing innovation with intellectual property rights remains essential. Legal professionals play a crucial role in navigating these issues to protect both creators and technology developers.
For those interested in further training on AI legal challenges and technologies, exploring comprehensive AI courses can provide valuable insights and practical skills. Resources like Complete AI Training’s latest AI courses offer relevant education tailored to legal and technical professionals.
Your membership also unlocks:
