Submission 261
Guardian Australia
Submitter group: Creative sector.
An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.
Summary
Read the original submission for the full text.
Guardian Australia says its parent group has AI licensing partnerships. Argues AI development should preserve copyright and voluntary licensing. Opposes text and data mining exemptions, safe harbours and mandatory licensing, and recommends opt-in consent, dataset transparency, enforceable rights reservations, crawler standards and accuracy and sourcing requirements.
Key points
Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.
Argues that growth of a domestic AI sector should not occur at the expense of established cultural and publishing industries by weakening existing intellectual property laws.
The growth of a domestic AI sector should not - and need not - occur at the expense of established cultural and publishing industries by weakening existing intellectual property laws.
Argues that a text and data mining exemption, expanded safe harbours or mandatory licensing would severely undermine the intent of the Copyright Act.
Introducing a TDM exemption, expanding safe-harbour provisions or forcing mandatory licensing would severely undermine the intent of the Copyright Act.
Argues that AI training on copyright material should require the rights holder's explicit prior authorisation, because opt-out models are inherently flawed.
Copyright-protected material should only be available for AI training or commercial ingestion with the rights holder’s explicit, prior authorisation. An opt-out model is inherently flawed because creators cannot track or restrict their work once it circulates on to third-party platforms beyond their direct control; a mandatory opt-in regime is the only practical safeguard.
Argues that the absence of training-data disclosure obligations, statutory weight for robots.txt and penalties for bypassing blocks makes rights protection ineffective.
AI developers face no obligation to disclose their training datasets; technical tools including robots.txt carry no statutory weight; and there are no penalties for companies that bypass bot-blocking measures.
Recommends conduct requirements for accuracy, integrity and source attribution of AI news answers, citing UK requirements for Google Search.
To stop Google from self-preferencing its own products over reliable journalism, the CMA mandated clear standards for output quality, factual accuracy and proper source attribution. Enforcing similar conduct requirements in Australia would help ensure AI providers are held accountable for the accuracy and integrity of their news substitutes, incentivising content deals with quality news sources.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Consult broadly on any proposed changes to copyright policy.
Any mandatory licensing scheme that sets a low floor for content payment, or safe-harbour exemptions that reduce the need for licensing, will directly impact the news industry and Australia’s information ecosystem. Therefore we support broad consultation on any proposed changes to copyright policy.
Explore voluntary collective licensing frameworks to bridge the complexity of licensing smaller creators, without granting broad legal exemptions to AI developers.
Voluntary collective licensing frameworks could be explored to bridge this gap but this should not be read as granting broad legal exemptions to AI developers.
Reject statutory text and data mining exemptions and safe-harbour provisions for AI training, and affirm the right to refuse use and receive fair compensation.
Explicitly reject statutory text and data-mining exemptions and refuse any safe-harbour provisions to cover AI training or content ingestion. Ensure that the copyright holder is understood to have the right to refuse the use of content as well as a requirement they receive fair financial compensation for the use of their content by AI chatbots.
Reject mandatory licensing schemes and keep voluntary commercial negotiation as the primary goal.
Reject any mandatory licensing schemes: This would strip publishers of their fundamental right to control their own intellectual property - specifically, the right to refuse use. Voluntary commercial negotiations should remain the primary goal.
Make copyright-protected material available for AI training or commercial ingestion only with the rights holder's explicit, prior authorisation (a mandatory opt-in regime).
Copyright-protected material should only be available for AI training or commercial ingestion with the rights holder’s explicit, prior authorisation. An opt-out model is inherently flawed because creators cannot track or restrict their work once it circulates on to third-party platforms beyond their direct control; a mandatory opt-in regime is the only practical safeguard.
Legislate preconditions for a licensing market: ensure an opt-out regime is not forced on creators, and require data-use transparency, statutory backing for opt-out mechanisms, and audit and enforcement mechanisms.
Ensure that an opt-out regime is not forced on creators. Support the market for licensed data by creating binding pre-conditions that mandate transparency about data use for training and grounding, give statutory backing to opt-out mechanisms, and create robust audit and enforcement mechanisms.
Create accuracy and sourcing requirements for AI platforms.
Create accuracy and sourcing requirements for AI platforms to protect Australia’s information ecosystem.
Require AI bots to disclose identity and purpose honestly, and prohibit bypassing technical access controls via false headers or residential proxies.
Establish minimum operational standards for AI bots to be honest about their identity and purpose. Make it illegal to bypass technical access controls via false headers or residential proxies.