Anthropic pushes for opt-out model for Australian content (www.theguardian.com)

Anthropic has proposed to the Australian government an opt-out model for copyrighted material, seeking conditional approval for AI firms to train their models on Australian content. This comes after the government ruled out a blanket copyright exemption, acknowledging concerns from public broadcasters like the ABC and SBS. They argue that AI companies should adhere to the same copyright and defamation laws as traditional media, warning that without strict regulations, AI poses a threat to public interest journalism by potentially "cannibalising" mainstream news. As part of its submission, Anthropic highlighted the dual potential of AI to boost Australia’s economy and tackle significant challenges while also acknowledging risks to national security and critical infrastructure resilience. The company suggested mechanisms like “do not crawl” signals to allow creators to control AI access to their works but faced skepticism from broadcasters who doubted the effectiveness of such measures. Both the ABC and SBS have called for stronger regulations that would require AI companies to negotiate fair media deals, emphasizing the necessity of protecting the integrity and sustainability of journalism in the face of growing AI technologies.