Anthropic and OpenAI have asked Australian lawmakers to reconsider rules that prevent them from training AI models on the country’s creative works. In submissions to a parliamentary inquiry, both companies accepted the government’s October 2025 decision not to create a broad text-and-data-mining exception in copyright law. That decision effectively requires developers to obtain permission from rights holders, limiting local training by most US-based labs. Anthropic proposed a narrow form of conditional approval for model training and said it would consider conditions such as investment or other measures that support Australian creators and cultural work. OpenAI called for a balanced framework that would let models learn from publicly available information while giving rights holders opportunities to collaborate. Both firms linked the issue to planned data-centre projects in Australia.
Anthropic and OpenAI have asked Australian lawmakers to reconsider rules that prevent them from training AI models on the country’s creative works. In submissions to a parliamentary inquiry, both companies accepted the government’s October 2025 decision not to create a broad text-and-data-mining exception in copyright law. That decision effectively requires developers to obtain permission from rights holders, limiting local training by most US-based labs. Anthropic proposed a narrow form of conditional approval for model training and said it would consider conditions such as investment or other measures that support Australian creators and cultural work. OpenAI called for a balanced framework that would let models learn from publicly available information while giving rights holders opportunities to collaborate. Both firms linked the issue to planned data-centre projects in Australia.