Back to Articles
Artists Risk Being Short-Changed by Australia’s AI Copyright Law

Crikey

ENRICHED

Details

Date Published
26 Aug 2026
Priority Score
2
Australian
Yes
Created
26 Aug 2026, 02:00 am

Authors (1)

Description

AI companies cannot use Australian books, art, music or other creative works to train AI without explicit consent, with heavy legal risks if they do. This could end up backfiring for artists.

Summary

This article examines the tension between Australia's restrictive AI copyright framework and its aspirations to attract major frontier AI developers like Anthropic. It highlights how the Treasury considers copyright 'clarity' a prerequisite for $21 billion in data centre infrastructure, suggesting that strict artist protections may conflict with national strategic goals for AI capability development. While primarily focused on intellectual property, the debate underscores the economic pressures that can influence the governance of frontier AI training environments and the sovereign control of AI compute resources.

Body

The debate over whether Australia should allow AI to train on scraped creative works has become entangled with a much larger prize: tens of billions of dollars in data centre investment and a strategic foothold in the emerging AI economy. US company Anthropic is reportedly looking to make Australia its “second home” for training AI models, with the associated infrastructure worth around US$15 billion (A$21 billion). But a Treasury briefing released in July showed Anthropic’s proposed expenditure was contingent on the “clarity” of Australia’s copyright law.