Submission 35
News and Media Research Centre, University of Canberra and News, Technology and Society Network, RMIT University
Submitter group: Universities.
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.
Argues that AI's integration into Australia's news and information environment needs measures on regional equity, representation of Australian content, copyright and licensing for news used in training, deepfake and synthetic media governance, and AI and media literacy. Makes ten recommendations on transparency, inclusion and public trust.
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 if AI systems favour large, metropolitan or commercially dominant sources, regional communities' local information may become invisible, and that equitable adoption and equitable visibility should be treated as connected.
If generative search systems and AI agents systematically favour large, metropolitan, English-language, freely crawlable or commercially dominant sources, communities outside major urban centres may find their local and regional information increasingly invisible. AI policy should treat equitable adoption and equitable visibility as connected issues.
Argues that the lack of transparency about the composition of training data for later models is a cause for concern, especially for culturally relevant Australian information.
Comparable language or geography level composition data have not been publicly disclosed for later models. The lack of transparency about data composition is a cause for concern, especially in the content of culturally relevant information for the Australian context.
States that Australian law is relatively clear that there is no copyright exemption for AI to use Australian media for training.
On the matter of using news content for training LLMs, Australian law is currently relatively clear on this. There is no copyright exemption for AI to use Australian media to train.
Argues that copyright alone cannot guarantee the continuing provision, plurality or accessibility of public-interest journalism.
Copyright may regulate the terms on which content is used, but it cannot by itself guarantee the continuing provision, plurality or accessibility of public-interest journalism.
Argues that public policy should extend AI and media literacy initiatives beyond formal schooling to families.
Public policy should extend AI and media literacy initiatives beyond formal schooling and ensure that families have access to practical, inclusive, and accessible education.
Argues that deepfake detection is limited by structural gaps, with key demographics missing from available visual and audio datasets.
The limitations are not only technical but structural: key demographics are missing from available visual and audio datasets. Audio training data is heavily concentrated on English and Chinese speech
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Support the production of high-quality regional news and information so AI systems have relevant, reliable content to access or train on, and improve regional media and AI literacy.
Support the production of high-quality regional news and information, so that AI systems have relevant and reliable content to access/train on, as well as improve the media and AI literacy of regional communities.
Require AI systems to assess and report news-source diversity, including regional, local, multicultural, First Nations, independent and specialist Australian media.
Require AI systems to assess and report on the diversity of news sources made visible through their services, including representation of regional, local, multicultural, First Nations, independent and specialist Australian media.
Ensure underrepresented communities contribute to the development of source data and training systems.
Ensure that underrepresented communities contribute to the development of source data and training systems so that they are more accurate and reflective of all communities.
Establish a funding stream for independently governed public-interest AI partnerships.
Establish a funding stream for independently governed public-interest AI partnerships.
Consider developing domain-specific public-interest AI capability through partnerships among public service media, news organisations, universities, libraries, archives, public institutions and communities.
The committee may also consider the possibility of developing domain-specific, public-interest AI capability through partnerships among public service media, news organisation, universities, libraries, archives, public institutions, and communities.
Enforce competition law as well as reform copyright to address AI interaction with news content.
This issue requires enforcing anti-trust law (competition law) as well as copyright reform.
Develop transparent partnership frameworks between news organisations and platforms, considering article links, authorship attribution, accurate reporting and return traffic.
There is an urgent need to develop frameworks of transparent partnerships between news organisations and platforms.18 Initial considerations may include whether the AI links the article, attributes the authorship, accurately represents the reporting and provides an opportunity for traffic to go back to the publisher.
Align Australia's approach to copyright law and licensing agreements with international precedents, particularly in the European Union.
Align Australia’s approach to copyright law and licensing agreements with international precedents, particularly in the European Union, to promote global consistency and legal certainty.
Apply fair-contribution mechanisms where AI companies derive substantial value from Australian news through training, retrieval, ranking, summarisation, answer generation or related AI-mediated uses.
Ensure fair-contribution mechanisms apply where AI companies derive substantial value from Australian news through training, retrieval, ranking, summarisation, answer generation or related AI-mediated uses.
Require transparency on how LLMs are trained and how AI platforms use Australian news content, and on licensing arrangements.
Require transparency regarding how LLMs are trained and how AI platforms generate content using Australian news content. In addition, require transparency in licensing arrangements to ensure the equitable treatment of news publishers in Australia and internationally.
Ensure development, testing and rollout of AI and media literacy education across the community, prioritising groups with high AI use or limited access to formal media literacy education.
Ensure the development, testing and rollout of AI and media literacy education across the community, with priority given to groups with high levels of AI use or limited access to formal media literacy education.
Extend AI and media literacy initiatives beyond formal schooling and ensure families have access to practical, inclusive and accessible education.
Public policy should extend AI and media literacy initiatives beyond formal schooling and ensure that families have access to practical, inclusive, and accessible education.
Develop a coordinated policy on provenance, digital signatures and verification standards for synthetic media, deepfakes and AI-generated disinformation.
there are significant regulatory gaps around synthetic media,29 deepfakes and AI-generated disinformation, which calls for a coordinated policy on provenance, digital signatures and verification standards.
Develop shared-accountability frameworks between journalists and AI developers, with clear disclosure.
Frameworks for shared accountability between journalists and AI developers and clear disclosure are needed.31
Support development of independent deepfake and AI detection datasets, with performance rigorously assessed across relevant population groups.
Support the development of independent deepfake and AI detection datasets and ensure that performance is rigorously assessed across relevant population groups.
Give regulators and qualified independent researchers secure, ongoing access to data needed to assess AI systems, enabling longitudinal auditing subject to privacy, security and commercial safeguards.
Provide regulators and qualified independent researchers with secure, ongoing access to data necessary to assess AI systems. This access should enable longitudinal auditing subject to appropriate privacy, security and commercial safeguards.