OpenAI and Anthropic are advocating for relaxed copyright restrictions in Australia to facilitate AI training, linking it to future investments.
OpenAI and Anthropic have urged the Australian government to ease restrictions on using local copyrighted content for training artificial intelligence models. This appeal was made during submissions to a parliamentary committee, as reported by Reuters.
Previously, Australia had rejected a proposal to create an exception that would permit developers to utilize such materials without needing separate licenses. Anthropic suggested a "narrow conditional approval" as a potential solution, which could involve commitments to support local authors and cultural initiatives.
OpenAI emphasized the need for a "balanced copyright regime" that would allow models to learn from publicly available data while still fostering collaboration with copyright holders.
Link Between Regulations and Investment Opportunities
The two companies highlighted the connection between copyright regulations and potential investments in Australia's AI infrastructure.
OpenAI has partnered with local operator NEXTDC to develop the S7 hyperscale data center project in Sydney, formalizing their collaboration with a memorandum. Meanwhile, Anthropic has secured its first data center lease in Queensland at the Zerra DC campus, which is set to deliver up to 2.16 GW of capacity by 2027, located about 250 km from Brisbane.
In its submission, OpenAI noted that Australia could significantly contribute to global AI infrastructure expansion, contingent on favorable copyright policies. However, the creative industries have voiced opposition to such exceptions.
Parliament Review Expected by Month-End
The Australian Parliament's Joint Committee on Artificial Intelligence is currently reviewing the effects of AI on national productivity and competitiveness, alongside its relationship with intellectual property laws. A key focus of this inquiry is the use of Australian creative and media content for AI training. Established on August 20, the committee is anticipated to release its final report by November 30.
This discussion comes amid ongoing debates regarding AI's economic impact. On September 22, Reserve Bank of Australia Governor Michele Bullock expressed skepticism about immediate productivity gains from AI technology, suggesting the current investment surge could resemble a bubble. She noted that the demand driven by data center construction might lead to inflationary pressures before any productivity improvements are realized.
Additionally, in July, a U.S. court approved a $1.5 billion settlement favoring Anthropic in a class-action lawsuit concerning the use of pirated book copies. The court had previously ruled that training models on legally obtained works constitutes fair use, distinguishing this from the illegal acquisition of materials.
