OpenAI’s Australian AI pitch comes with a copyright catch

OpenAI has linked any Australian AI model-training infrastructure to changes in copyright law, saying the current settings are a blocking issue before it will discuss the energy requirements of a local training centre.

In an interview with The Australian, OpenAI vice president of global policy Ann O’Leary said the company could not proceed with model training in Australia under the existing framework.

“We will not be able to build a … training centre here if the copyright law is as it exists,” she said. “Our company isn’t ready to have that conversation if we can’t get through the gating problem of the copyright barriers.”

READ MORE: Albanese government rejects push to relax copyright laws for AI training

READ MORE: Anthropic inks AI pact with Albanese government & preps Sydney office

OpenAI runs into Labor’s copyright line

The position puts OpenAI directly at odds with a key part of the Albanese government’s developing AI policy.

In its July AI framework, the government said Australian writers, artists and journalists should retain ownership of their work and that companies should not use Australian creative works to train AI without the creator’s control.

The government has also formally ruled out a text and data mining exception to copyright law. Its Copyright and AI Reference Group is instead examining issues including licensing arrangements, the legal status of AI-generated material and cheaper avenues for copyright enforcement.

O’Leary suggested changing those settings could put Australia at the front of OpenAI’s international infrastructure expansion.

“No other place in the world has done (AI training) this yet, and so the question is: is Australia going to be the first? And that’s really a copyright question more than a renewable energy question,” she said.

Energy is the next hurdle

Copyright is only one of the conditions facing large AI developments in Australia.

The government’s proposed standards would require large data centres to underwrite new power supply, cover their share of grid connection costs, reduce demand when required and meet water-efficiency requirements.

O’Leary’s comments effectively put those infrastructure negotiations behind the copyright question: OpenAI says it is not ready to progress the former until it has a legal pathway for training models locally.

The size of the investment opportunity helps explain the government’s interest. Australian Bureau of Statistics figures show information media and telecommunications businesses invested about $8.7 billion in buildings and equipment during the March quarter, with the ABS saying data centre equipment drove a significant part of the increase.

The Productivity Commission has separately estimated AI could add about $116 billion to Australian economic activity over the next decade, although it has stressed there is considerable uncertainty around the estimate.

Assistant Minister for Science, Technology and the Digital Economy Andrew Charlton told The Australian the economic impact could rival the mining boom, while arguing AI’s productivity effects could spread more broadly across businesses and industries.

Copyright deals already exist

The dispute comes as AI companies and rights holders increasingly negotiate commercial agreements rather than waiting for copyright law to be rewritten.

OpenAI itself has struck several publisher partnerships. Its multi-year agreement with News Corp gives it access to current and archived journalism from mastheads including The Australian, news.com.au, The Daily Telegraph and Herald Sun, with permission to display content in responses and use it to enhance OpenAI products.

That agreement does not resolve the separate legal question of what material can be used to train frontier AI models in Australia. But it highlights the central divide in the policy debate: creative industries are pushing licensing and consent as the path forward, while OpenAI says Australia’s existing legal settings remain too restrictive for local model training.

With the government preparing legislation for its broader AI standards, OpenAI’s intervention turns that copyright dispute into an investment question as well as a fight over creators’ rights.