Anthropic urges Australia to consider "conditional approval" for Big Tech to train models on copyrighted works, giving copyright holders a choice to opt out (Josh Butler/The Guardian)

Anthropic has called on the Australian government to adopt a “conditional approval” framework that would allow large-tech firms to train AI models on copyrighted material while giving copyright holders the option to opt out. The proposal seeks to balance AI development with creators’ rights by enabling selective use of protected works. It aims to provide a legal pathway for training data that respects intellectual-property interests.