OpenAI and Anthropic Push Australia to Ease AI Training Ban

OpenAI and Anthropic Challenge Australia's Copyright Stance

OpenAI and Anthropic have formally requested that Australia reconsider its ban on using copyrighted Australian content to train AI models, proposing a middle-ground approach that would tie limited copyright exemptions to infrastructure investments and programs supporting local creators.

In submissions filed in September to a parliamentary inquiry examining AI regulations, both companies acknowledged Australia's firm position against broad copyright exemptions but argued for a more nuanced framework. Anthropic, which operates the Claude chatbot, proposed "a narrow form of conditional approval" that would permit AI training under specific circumstances designed to benefit Australian artists and cultural institutions.

OpenAI similarly called for "a balanced copyright framework that allows models to learn from publicly available information" while providing rights holders with collaboration opportunities. The company emphasized that such a framework could coexist with protections for content creators.

Linking Policy to Infrastructure Investment

Both companies connected their policy requests to substantial data center commitments in Australia. OpenAI has secured an offtake agreement with Australian data center operator NextDC for a planned Sydney facility. Anthropic was recently announced as a partner for a data center project in Queensland.

The timing of these submissions places additional pressure on Australian policymakers, who are attempting to balance competing interests. On one side are AI developers seeking access to training data and regulatory clarity. On the other are authors, publishers, and creative professionals who have consistently opposed allowing their work to be used for commercial AI training without explicit consent or compensation.

Why it matters

Australia's decision will influence how other nations approach the fundamental tension between AI development and intellectual property rights. As countries compete for hundreds of billions in projected AI investment through the end of the decade, the framework Australia establishes could become a reference point for balancing innovation incentives with creator protections—particularly for mid-sized economies seeking to attract tech investment without sacrificing domestic creative industries.

Regulatory Timeline and Broader Context

Australia has announced plans to implement comprehensive AI and data center regulations beginning next year. The government has already indicated it will not provide the type of broad copyright exemptions that AI companies initially sought, which would have allowed unrestricted use of copyrighted material for model training.

The Joint Select Committee on Artificial Intelligence, which received the companies' submissions, is scheduled to deliver its report in November. That timeline suggests Australia's policy direction on AI training and copyright will become clear within months.

The debate reflects a global challenge as governments worldwide attempt to craft AI regulations that encourage technological development while addressing concerns about data usage, creator rights, and economic impact. Economists project the AI sector could generate hundreds of billions of dollars in value by decade's end, making regulatory decisions particularly consequential for national competitiveness.

These details were first reported by AI Watch.