OpenAI and Anthropic, two leading artificial intelligence companies, are urging the Australian government to reconsider its stance on using local creative content for AI model training. The companies believe that a limited exemption from copyright restrictions could strike a balance between protecting artists’ rights and fostering significant investment in the burgeoning AI sector.
A Delicate Balancing Act
As artificial intelligence continues its rapid advancement, governments worldwide are scrambling to establish regulatory frameworks while simultaneously vying for a substantial share of AI investments, which economists predict could reach hundreds of billions of dollars by the end of the 2020s. Australia is set to implement a comprehensive set of AI and data center regulations starting next year. Crucially, the government has so far ruled out the possibility of including exemptions within its copyright law.
Under the current regulations, AI firms, many of which are registered in the United States, are effectively barred from using copyrighted local content to train their AI models in Australia. In submissions to a parliamentary inquiry into AI law in September, both OpenAI and Anthropic acknowledged the existence of these restrictions but continued to advocate for a relaxation of the current limitations.
Proposed Solutions
Anthropic, in its submission, stated, “We accept that broad copyright exemptions are not universally supported and that the government has ruled out such an approach. As an alternative, we believe the government could allow for AI model training through a narrowly scoped conditional approval mechanism.” The company further added, “We would be open to considering conditions proposed by Australia that are oriented towards investment or other initiatives designed to support the future development of its creators and cultural industries.”
OpenAI, on its part, called for the establishment of a balanced copyright framework. This framework, they suggest, should enable models to learn from publicly available information while also providing opportunities for rights holders to collaborate with AI companies. This approach aims to ensure that creators are appropriately recognized and potentially compensated for the use of their work in AI development.
AI Investment and Copyright Policy
Both companies have explicitly linked their copyright policy suggestions to substantial investments in Australia’s large-scale AI infrastructure. OpenAI has already signed a power purchase agreement with NextDC, an Australian data center developer, for a planned data center in Sydney. Similarly, Anthropic recently partnered with a data center in Queensland. These moves underscore the AI companies’ commitment to establishing a physical presence and investing in the country’s technological backbone.
Navigating Creator Concerns
The input from OpenAI and Anthropic adds further pressure on the Australian government. Authorities are now faced with the complex task of weighing the interests of AI developers against those of authors, publishers, and other creators. These creative professionals have consistently voiced strong opposition to allowing companies to use their work for commercial AI model training without explicit permission or payment.
The government’s decision will have significant implications for the future of AI development in Australia, potentially influencing the country’s competitiveness in the global AI race and its ability to attract further innovation and investment. The ongoing debate highlights the global challenge of adapting intellectual property laws to the rapid advancements in AI technology.
