Company Updates

Anthropic and OpenAI push for Australian copyright reform to train AI models

Share
Anthropic and OpenAI push for Australian copyright reform to train AI models

Anthropic and OpenAI have formally requested that Australian lawmakers reconsider copyright laws that currently prohibit AI training on local creative content without explicit permission. Both companies tie their requests to planned data center investments in the country.

TL;DR

  • Anthropic and OpenAI seek Australian copyright law reforms to permit AI training on local creative works.
  • Both companies link their requests to planned data center investments in Australia.
  • The request comes as Australia prepares to introduce broader AI and data center regulations next year.

What happened

Anthropic and OpenAI have submitted requests to a parliamentary inquiry, asking Australian lawmakers to reconsider copyright laws that currently require permission from copyright holders for AI training on local creative works. The companies acknowledge the government's decision to exclude a text and data mining exception but propose alternative solutions.

Anthropic suggests a 'narrow form of conditional approval' for model training, with potential conditions like investment in Australian creators. OpenAI requests a balanced framework that allows models to learn from publicly available information while collaborating with rights holders.

Both companies highlight their planned data center investments in Australia. OpenAI has a power purchase agreement with NextDC for a Sydney data center, while Anthropic is partnering on a Queensland data center project. Anthropic's CEO, Dario Amodei, previously raised the copyright issue with Australian Treasurer Jim Chalmers, stating that clarity on copyright settings is crucial for their investment.

Why it matters

This development could significantly impact AI startups and investors, as it may set a precedent for how countries balance copyright protection with AI innovation. If successful, it could open up new opportunities for AI training and investment in Australia.

The outcome may also influence global AI policy, as other countries grapple with similar issues regarding AI training and copyright. For developers, this could mean easier access to diverse datasets, while for investors, it could signal new markets and opportunities.

However, the creative industries in Australia have strongly opposed these changes, arguing that AI companies should compensate copyright holders, as in the case of Anthropic's $1.5 billion settlement in the US over unauthorized use of books in its training data.

Key facts

  • Anthropic and OpenAI submitted requests to a parliamentary inquiry seeking changes to Australian copyright laws for AI training.
  • Anthropic proposes a 'narrow form of conditional approval' for model training with potential investment conditions.
  • OpenAI requests a balanced framework for AI learning from publicly available information while collaborating with rights holders.
  • Both companies tie their requests to planned data center investments in Australia: OpenAI in Sydney and Anthropic in Queensland.
  • Anthropic's CEO, Dario Amodei, previously raised the copyright issue with Australian Treasurer Jim Chalmers.
  • Australia intends to introduce broader AI and data center regulations next year.
  • Anthropic agreed to pay $1.5 billion to settle a class action in the US over unauthorized use of books in its training data.
  • The submissions were made to a joint select committee on AI established by both houses of Australia's parliament.

Context

This development comes as Australia prepares to introduce more comprehensive AI and data center regulations next year. The outcome of these discussions could set a precedent for how countries balance the needs of AI innovation with the protection of creative industries.

Globally, the issue of AI training on copyrighted material is a contentious one. In the US, courts have ruled that training on lawfully acquired books could be considered fair use, but this does not extend to pirated material, as seen in Anthropic's settlement.

For AI startups and investors, the outcome of these discussions could open up new markets and opportunities, while also influencing global AI policy. However, the strong opposition from creative industries highlights the need for a balanced approach that respects copyright holders' rights.

Topics

Related coverage

Join the discussion

Have a take on this story? Weigh in with our community on Facebook.

💬 Discuss on Facebook →