OpenAI and Anthropic urge australia to relax AI training content restrictions

On September 22, 2026, Reuters reported that OpenAI and Anthropic have called on the Australian government to reconsider its ban on using locally protected creative content to train AI models. The companies argue that limited exemptions could protect artists' rights while attracting more investment. Australia plans to implement a comprehensive AI and data center regulatory framework starting in 2027, excluding copyright law exemptions. Currently, these U.S.-based firms cannot use Australian copyrighted content for model training. In September 2026, both companies acknowledged the ban in parliamentary AI law submissions but requested relaxed restrictions. Anthropic suggested a conditional approval mechanism for AI training, contingent on Australia offering investment or support for local creators. OpenAI called for a balanced copyright framework allowing models to learn from public information while enabling collaboration with rights holders. Both companies also linked copyright policies to Australia’s large AI infrastructure investments, with OpenAI partnering with NextDC for a Sydney data center and Anthropic collaborating with Queensland on a data center project. These proposals have increased pressure on the Australian government to balance AI developers’ interests with creators’ concerns over unauthorized use of their work.