Anthropic and OpenAI have called on Australia to relax its ban on using domestic creative content to train artificial intelligence models, telling a parliamentary inquiry that a narrow conditional exemption could protect creators while encouraging investment. The submissions, lodged in September 2026 with the Joint Select Committee on Artificial Intelligence, mark a push by OpenAI and Anthropic to reshape Australian copyright policy as the government finalises AI and data centre rules due to take effect next year, Reuters reported.
Key points
- Anthropic proposed a “narrow form of conditional approval” for AI model training and said it would consider investment requirements to support Australian creators.
- OpenAI advocated a “balanced copyright framework that allows models to learn from publicly available information” with collaboration opportunities for rightsholders.
- Australia has already ruled out broad copyright exemptions that would allow mostly US-domiciled companies to train models domestically.
- OpenAI has signed an offtake agreement with NextDC for a Sydney data centre; Anthropic was named last week as a partner for a Queensland data centre.
- The Joint Select Committee on Artificial Intelligence is due to deliver its report in November.
Conditional approval as alternative to broad exemption
Anthropic acknowledged in its submission that broad exceptions from copyright law “have not received widespread support and have been ruled out by the government.” Instead, the company argued the government “could permit AI model training through a narrow form of conditional approval.” It added that if Australia required “investment or other conditions designed to support the future of Australian creators and cultural endeavours … we are open to considering those conditions.” The formulation shifts the debate from a blanket exemption to a negotiated framework in which training rights are contingent on measurable commitments to the local creative sector.
OpenAI links framework to infrastructure commitments
OpenAI’s submission called for “a balanced copyright framework that allows models to learn from publicly available information” while offering rightsholders opportunities to collaborate. The company tied its position to concrete infrastructure investment: it has signed an offtake agreement with Australian data centre developer NextDC for a planned facility in Sydney. Anthropic, meanwhile, was confirmed last week as a partner for a data centre in Queensland. Both submissions frame the training question as inseparable from the capital the labs are already deploying on Australian soil.
Government position and creator opposition
Australia plans a broad set of rules on AI and data infrastructure to take effect from next year, yet has already rejected a copyright exemption that would effectively allow the mostly US-domiciled companies to train models domestically. Authors, publishers and other creators have resisted exemptions allowing their work to be used for commercial model training without consent or payment. The Joint Select Committee on Artificial Intelligence is due to deliver its report in November.
