The Pulse
Anthropic Seeks Conditional Copyright Approval in Australia
Anthropic has asked Australia to consider a conditional opt-out framework for AI training on copyrighted works, while the ABC and SBS have urged stronger rules for AI companies. The government has ruled out a broad text and data mining exce
AI.info Team · Written with AI ·
“The AI giant Anthropic has urged the Albanese government to consider giving ‘conditional approval’ for big tech to train its models on Australian copyrighted works under an opt-out model, after conceding it won’t secure a blanket copyright exemption.”
Josh Butler, The Guardian reporter, writes
Anthropic proposes a narrower route
Anthropic is asking Australia to consider allowing AI developers to train models on copyrighted material through what the company calls a “narrow form of conditional approval.” In its submission to a federal parliamentary inquiry, Anthropic acknowledged that a broad text and data mining exception has been ruled out by the government and has not received widespread support.
The company said Australia should provide legal certainty that brings AI developers to the table while advancing national cultural interests. It suggested that “do not crawl” signals in robots.txt files could be part of the approach. These signals tell web crawlers which material they may access. Anthropic also said it was open to supporting Australian creators. The proposal is a request for lawmakers to consider a framework; it is not an approval already granted by the government.
Anthropic presented both potential benefits and risks from AI development in its submission. It said the technology could transform the Australian economy, increase productivity and help address national challenges, while warning of possible effects on national security, critical infrastructure and public safety. The company also promoted plans to invest billions of dollars in Australia, including in data centres, and called on the government to support frontier model training.
Government rules out a broad exception
The government has ruled out a text and data mining exception that would let AI developers use Australian creators’ works for training without permission or payment. In an October 2025 release, Attorney-General Michelle Rowland said the government was consulting on possible copyright-law updates, but that the work would not include such an exception.
The government said it was examining legal routes for using copyrighted material in AI, including whether to create a paid collective licensing framework or retain voluntary licensing. Rowland said Australian creatives should benefit from AI opportunities and that the government had no plans to weaken copyright protections. The release also identified clearer rules for material generated using AI and less costly enforcement as areas under consideration.
Broadcasters challenge the opt-out approach
The ABC called for AI companies to face the same copyright, defamation and privacy rules as media outlets. In its submission, the broadcaster argued that AI services currently receive more favourable treatment than established Australian media and warned that the unequal treatment could accelerate what it called the “cannibalisation” of the news industry.
SBS questioned whether an opt-out mechanism would work in practice. It said robots.txt mechanisms are routinely ignored or bypassed, and argued that requiring rights holders to monitor and update protections places too much responsibility on them. SBS said AI companies should bear the onus of avoiding wrongdoing.
Both broadcasters suggested including AI companies in the news bargaining incentive. Under the incentive, platforms that do not enter or renew commercial agreements with news publishers would pay a charge; platforms with qualifying agreements could offset their liability. The incentive applies to large digital platforms operating significant social media or search services, whether or not they carry news.
Inquiry examines copyright and AI
The Joint Select Committee on Artificial Intelligence is examining how AI interacts with existing intellectual property and copyright laws, including the use of Australian creative, cultural and media content to train models. Its remit also covers AI’s opportunities and risks more broadly. Parliament lists 30 November 2026 as the committee’s reporting date.
The dispute brings two competing concerns before the inquiry: Anthropic’s request for a legally defined route to train on copyrighted material, and broadcasters’ warnings about the effects of AI services on creators and public-interest journalism. The government has rejected a broad training exception, but its consultation is considering other copyright approaches.