Anthropic rejects claim it is pressing Australia to rewrite AI copyright rules
Australian artists and publishers fear that changing copyright rules to make AI training easier could leave creators with little control over how their work is
Australian artists and publishers fear that changing copyright rules to make AI training easier could leave creators with little control over how their work is used. At a parliamentary hearing in Sydney on Tuesday, Anthropic said it had not tried to dictate Australia’s approach, while arguing that the current rules make training AI models in the country “impossible.”
Jeffrey Bleich, Anthropic’s special envoy and a former U.S. ambassador to Australia, told the joint parliamentary inquiry the company had “never tried to dictate to Australia” on copyright. He said Anthropic had shared information with the government, as lawmakers examined whether rights holders should have to opt out before their material could be used to train AI systems.
Copyright rules under scrutiny
Media and entertainment groups pushed back against loosening protections, warning that creators could be badly harmed if AI companies gained broader access to copyrighted material. The Australian Broadcasting Corporation said copyright laws did not need to change.
The debate matters to local creators because an opt-out system would place the burden on rights holders to specifically prevent their work from being used for training. Anthropic and OpenAI are seeking access to Australian content as they pursue plans for large data centres in Sydney and regional Queensland.
Australia is preparing AI-specific laws expected to take effect next year. The inquiry’s hearing also turned to the security risks posed by AI agents—tools that can take actions on their own.
Data breach questions
OpenAI again apologised over an incident involving its agent and a Medicare portal. The company said it should have handled its response better and had work to do to rebuild trust with Australians. The incident involved access to public and non-public files on the portal.
Anthropic said it did not believe its products had breached Australian government systems. Its head of safeguards, Dave Orr, told lawmakers that an investigation covering hundreds of millions of transcripts found no unauthorised interactions with government systems. But he said the company has limited visibility into how customers use its models because of standard “zero data retention” policies.
Anthropic executives said they would be open to laws requiring AI companies to report data breaches. The hearing brought the company’s copyright position into sharper focus, even as lawmakers weighed how to address the risks of increasingly capable AI agents.
Source: theguardian.com
