Anthropic has asked the Albanese administration to consider a narrow, conditional-approval framework that would let big-tech firms train models on Australian copyrighted material only if rights holders opt in, after acknowledging that a blanket exemption will not be granted. The company frames the proposal as a way to give legal certainty while respecting cultural interests.
Australia’s public broadcasters, the ABC and SBS, have responded with strong criticism, urging the government to impose robust new rules that would require AI companies to comply with the same copyright, defamation and privacy standards that apply to traditional media. Both outlets warned that unchecked AI use could erode the financial base of mainstream journalism.
In its submission, the ABC warned that “as publishers, AI-based services are currently unregulated and receive more favourable treatment than established Australian media. This unequal treatment will accelerate the cannibalisation of the Australian news industry,” emphasizing the perceived competitive imbalance.
SBS added that large language models such as ChatGPT or Claude pose “challenges to the sustainability of the Australian news industry and the erosion of information integrity and public trust through their use of news content without permission, attribution or compensation,” highlighting concerns over content misuse.
The joint select committee on artificial intelligence will hold hearings in Sydney next week, with representatives from Anthropic and OpenAI slated to appear after OpenAI disclosed that one of its agents accessed Services Australia data in June, underscoring regulatory scrutiny of AI data practices.
Anthropic’s filing highlighted potential economic benefits, describing AI as a catalyst for productivity gains and a tool for tackling national challenges, while also noting “serious implications for national security, critical infrastructure resilience and public safety.” The company signalled plans to invest billions in Australian data centres and other infrastructure to support “frontier model training.”
The Labor government has ruled out a text-and-data-mining exemption, with Prime Minister Anthony Albanese pledging “the strongest possible protection” for Australian creators. Nonetheless, the idea of an opt-out mechanism, allowing rights holders to decide whether their works can be used for AI training, remains under discussion.
Anthropic proposed that the government allow model training through a narrow conditional-approval process, using “do not crawl” signals in robots.txt files to respect opt-out choices. The ABC countered that such mechanisms are often ignored, placing the monitoring burden on rights holders rather than on AI developers, and called the current system “anti-competitive.”