ABC Confirms AI Scraping, Rejects Opt-Out Model Amid Australian Copyright Debates

The Australian Broadcasting Corporation (ABC) has told a parliamentary inquiry that AI companies have likely used its content to train models, but it opposes an opt-out model for content usage. Instead, ABC supports licensing deals, aligning with Australia’s government, which ruled out broad copyright exceptions for AI training in October 2025. The debates reflect broader tensions over data scraping, fair compensation, and regulatory frameworks.

ABC Confirms Likely AI Scraping, Opposes Opt-Out Model

Kate Gilchrist and David Sutton, representing the Australian Broadcasting Corporation (ABC), told a parliamentary inquiry that AI companies have likely scraped the broadcaster’s content to train their models. This claim aligns with broader concerns about how AI firms source training data from publicly available material.

While ABC acknowledges the likelihood of scraping, it does not support an opt-out model for content usage. Instead, the broadcaster favors licensing deals, arguing that current copyright laws already provide a framework for AI firms to legally access content—provided they pay for it.

Gilchrist and Sutton emphasized that an opt-out model would unfairly shift the burden onto rights holders, requiring them to actively monitor and restrict AI companies’ access to their work.

Australia’s Regulatory Landscape and AI Industry Pressure

Australia currently has at least four state and federal inquiries examining AI’s impact on copyright, data scraping, and regulatory policies. These inquiries reflect growing tensions between media organizations, tech companies, and policymakers over how to balance innovation with fair compensation.

In October 2025, Australia’s government ruled out a broad copyright exception that would allow AI firms to train on copyrighted material without permission or payment. This decision followed lobbying efforts from companies like OpenAI and Anthropic, which have argued that restrictive copyright laws limit their ability to train models on local content.

David Orr, Head of Safeguards at Anthropic, told the inquiry that the company had found no evidence of its products breaching Australian government systems. This statement followed a investigation into an OpenAI agent’s alleged hack via Hugging Face, which also yielded no proof of unauthorized access.

Background: Global Context on AI and Copyright

Regulatory scrutiny of AI training practices is not unique to Australia. In the European Union, OpenAI recently introduced an invisible watermarking system for text and code generated within the bloc, a measure designed to comply with the EU’s AI Act. However, the effectiveness of such tools remains limited, particularly for shorter or edited outputs.

Beyond copyright, AI ethics and safety have become flashpoints globally. In the U.S., a former researcher at Anthropic and OpenAI is set to testify about the risks of advanced AI systems, while the Pentagon has cut ties with Anthropic over concerns about supply chain security. These developments underscore the complex trade-offs between innovation, regulation, and national interests.

What this means

Lazyfounder analysis — our interpretation, not reported fact.

Australia’s stance on AI and copyright is shaping up as a microcosm of global debates. Unlike jurisdictions pushing for stricter controls—such as the EU with its AI Act—Australia is leaning toward a market-driven approach, where licensing negotiations determine access to training data. This positions media organizations like ABC as gatekeepers, but it also risks creating a fragmented landscape where only the largest players can afford to participate in licensing deals.

For startups and AI operators, the message is clear: Australia is unlikely to offer broad exceptions for data scraping, and reliance on opt-out mechanisms is not a viable strategy. The emphasis on licensing could advantage incumbents like OpenAI and Anthropic, which have the resources to negotiate deals, while smaller firms may struggle to compete. The ongoing inquiries could still shift the balance, but for now, the status quo favors rights holders over unrestricted access.

The broader takeaway is that AI regulation is not just a technical or legal challenge—it’s a negotiation between industries with competing priorities. Media companies want compensation, AI firms want access, and governments are caught in the middle. How Australia resolves this tension may set a precedent for other countries grappling with the same questions.

Key takeaways

• AI companies have likely used Australian media content, including ABC’s, for training without explicit permission.
• ABC prefers licensing deals over opt-out models, placing the onus on AI firms to negotiate access.
• Australia’s government has rejected blanket copyright exceptions for AI training but is grappling with regulatory approaches.
• Debates over AI ethics, data scraping, and copyright remain unresolved as multiple inquiries continue.
• Anthropic and OpenAI are pushing for relaxed copyright rules in Australia, citing barriers to training on local content.

FAQ

Why does ABC oppose an opt-out model for AI training?

ABC argues that an opt-out model places the burden on rights holders to monitor and restrict AI companies’ use of their content. Instead, it supports licensing deals, which ensure compensation and clearer legal frameworks.

What is Australia’s current position on copyright exceptions for AI training?

In October 2025, the Australian government ruled out a broad copyright exception that would allow AI firms to train on copyrighted material without permission or payment.

How are OpenAI and Anthropic responding to Australia’s copyright rules?

Both companies have lobbied for relaxed copyright rules, arguing that current laws limit their ability to train models on Australian content. They have also emphasized their compliance with security and data protection standards.

What are the implications of ABC’s stance for AI startups?

ABC’s preference for licensing deals means that startups may face higher costs and barriers to accessing training data. Smaller firms without the resources to negotiate deals could be at a disadvantage.

How does Australia’s approach compare to other countries?

Unlike the EU, which has introduced strict regulations like the AI Act, Australia is pursuing a market-driven approach. This relies on licensing negotiations rather than broad legal exceptions or opt-out mechanisms.

Related on Lazyfounder

• OpenAI
• Anthropic
• Hugging Face
• More ai news
• OpenAI to add invisible watermarks to EU-generated text for AI Act compliance