AI developer Anthropic has urged the Australian government to give conditional approval for training AI models on copyrighted works using an opt-out framework. The request follows the Labor government’s decision to rule out a blanket text and data mining exception. In its submission to a parliamentary inquiry, Anthropic claimed frontier model training could boost economic productivity and offered to invest billions in local data centers.
However, Australian public broadcasters ABC and SBS strongly opposed broad exemptions, urging strict regulation and inclusion of AI companies in news bargaining frameworks. The ABC warned that unregulated AI usage threatens to cannibalize mainstream journalism, while SBS stated that unauthorized training erodes public trust and news sustainability.
Why it matters
Australia’s policy decisions on opt-out copyright systems could establish key global legal precedents for AI training data rights.
Media publishers are pushing to force AI labs into commercial news bargaining frameworks to secure training compensation.
Source: theguardian.com



