Submission 274
QUT Centre for Data Science
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.
Describes seven key messages from a university data science centre, with numbered recommendations: sovereign AI capability and data sovereignty, responsible AI as competitive advantage, data governance, IP and copyright modernisation, risk resilience, workforce capability and value-based adoption measurement. Cautions against assuming that every contributing author or centre member subscribes to every comment or recommendation.
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 Australia should ensure AI infrastructure investment creates a chain from Australian resources to AI infrastructure, access, capability and public benefit.
Australia should ensure that this investment creates a chain from Australian resources, to AI infrastructure, to Australian access, to Australian capability, and ultimately to Australian public benefit.
Argues that AI governance should be proportionate and risk-based, focusing regulatory effort on higher-risk applications while avoiding unnecessary restrictions on low-risk uses.
AI governance should be proportionate and risk-based, avoiding unnecessary restrictions on low-risk uses that could inhibit innovation, experimentation and productivity gains. Regulatory effort should focus on higher-risk applications, harmful conduct and demonstrable harms, supported by clear accountability, enforcement and avenues for redress.
Argues that strong data governance, stewardship and Indigenous Data Sovereignty are foundations for trustworthy AI.
The Centre's work emphasises responsible data stewardship, FAIR data principles, privacy-preserving analytics, and Indigenous Data Sovereignty as foundations for trustworthy AI systems.
Argues for a statutory text and data mining exception limited to non-commercial academic research, with safeguards against commercial use.
A statutory exemption in line with international best practice would legally permit researchers to mine copyrighted datasets solely for non-commercial, academic research purposes, ensuring Australian universities remain globally competitive without infringing on creator rights. Such an exemption needs to ensure that commercial organisations cannot exploit this as a loophole to bypass protections for authors and artists.
States that, while AI offers significant opportunities, there are several risks requiring proactive management, and that policy should address emerging harms while supporting innovation.
While AI offers significant opportunities for Australia, there are several risks that require proactive management, and policy should seek to address and mitigate these emerging harms while supporting innovation and responsible AI deployment.
Argues tool access alone does not establish effective organisational AI use or measurable value.
Access to AI tools, licences, infrastructure or models does not necessarily indicate that AI has been successfully embedded into organisational practices or that it is producing measurable value.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Expand the National AI Plan 2025 through a dedicated Sovereign AI Capability Framework linking Australian resources, compute, data, research, workforce, innovation and commercialisation to Australian public benefit outcomes.
Expand the National AI Plan 2025 [5] through a dedicated Sovereign AI Capability Framework that links Australian resources, compute infrastructure, data assets, research capability, workforce development, innovation, and commercialisation to clear Australian public benefit outcomes.
Support co-design and funding of sovereign AI and data infrastructure, including Australian-hosted AI models and strategically important datasets, with consideration of energy, water and environmental sustainability. Include long-term national capability needs.
Support the co-design and funding of sovereign AI and data infrastructure, including Australian-hosted AI models and strategically important datasets, while ensuring appropriate consideration of energy, water, environmental sustainability, and long-term national capability needs.
Retain human decision-makers in high-risk or consequential AI settings and provide clear human-review avenues.
AI should not fully replace human decision-makers in high-risk or consequential settings, and clear avenues for human review should be available.
Build on existing governance through a national Responsible AI Framework extending beyond government to industry, research institutions, public-interest organisations and community applications.
Build on Australia's existing AI governance arrangements by developing a nationally consistent Responsible AI Framework that extends beyond government to support industry, research institutions, public-interest organisations, and community applications.
Embed transparency, explainability, human oversight, accountability, Indigenous Data Sovereignty, CARE principles and review and redress throughout the AI lifecycle.
The framework should embed transparency, explainability, human oversight, accountability, Indigenous Data Sovereignty, the CARE principles and mechanisms for review and redress across the AI lifecycle.
Make governance risk-based, enabling low-risk innovation and productivity while applying stronger oversight, enforcement and redress where risks or harms are greater.
Governance should be risk-based, enabling low-risk innovation and productivity gains while applying stronger oversight, enforcement and redress where risks or harms are greater.
Include proportionate risk classification, AI registers, accountable ownership, lifecycle assurance, monitoring, traceability, human review and challenge and redress mechanisms.
This should include proportionate risk classification, AI system registers, accountable ownership, lifecycle assurance, monitoring, traceability, human review and mechanisms for challenge and redress.
Strengthen Indigenous Data Sovereignty, AI literacy and public participation through meaningful Aboriginal and Torres Strait Islander involvement and capability for individuals, organisations and communities to critically engage with AI and its impacts.
Strengthen Indigenous Data Sovereignty [7-10], AI literacy, and public participation in AI governance by ensuring meaningful involvement of Aboriginal and Torres Strait Islander peoples and building the capability of individuals, organisations, and communities to critically engage with AI systems and their impacts.
Develop a National AI Data Governance Framework using government data-governance and AI-assurance frameworks, strengthening discoverability, access, quality, interoperability, metadata, ownership, stewardship, provenance, privacy, accountability and responsible training-data use across government, industry, research and public-interest applications.
Recommendation 3.1: Develop a National AI Data Governance Framework that builds on the Australian Government Data Governance Framework [11] and the National Framework for the Assurance of Artificial Intelligence in Government [13] to strengthen data discoverability, accessibility, quality, interoperability, metadata, ownership and stewardship, provenance, privacy protection, accountability, and the responsible use of training data across government, industry, research organisations, and public-interest applications.
Embed Indigenous Data Sovereignty and Governance using Maiam nayri Wingara, CARE and Commonwealth frameworks, with meaningful Aboriginal and Torres Strait Islander leadership and decisions over Indigenous data and knowledge governance, access, use, sharing and stewardship.
Recommendation 3.2: Embed Indigenous Data Sovereignty and Indigenous Data Governance within Australia's AI data ecosystems through adoption of the Maiam nayri Wingara Indigenous Data Sovereignty principles, the CARE Principles for Indigenous Data Governance, and the Commonwealth's Framework for Governance of Indigenous Data, supported by meaningful Aboriginal and Torres Strait Islander leadership and decision-making in the governance, access, use, sharing, and stewardship of Indigenous data and knowledge.
Consider amending the Acts to establish protection thresholds for AI-assisted works and the parameters for measuring them.
The Committee should consider amending the Acts to similarly establish a threshold for when AI-assisted works are no longer afforded protections, and the parameters by which these thresholds are measured.
Introduce a statutory text and data mining exemption permitting researchers to mine copyrighted datasets solely for non-commercial academic research, with safeguards so commercial organisations cannot use it to bypass protections for authors and artists.
A statutory exemption in line with international best practice would legally permit researchers to mine copyrighted datasets solely for non-commercial, academic research purposes, ensuring Australian universities remain globally competitive without infringing on creator rights. Such an exemption needs to ensure that commercial organisations cannot exploit this as a loophole to bypass protections for authors and artists.
Require transparency and disclosure of Australian content used in AI training, acknowledging Indigenous Cultural and Intellectual Property rights; provide appropriate licensing or remuneration for creators.
Transparency and disclosure requirements are necessary to ensure that Australian rights holders know when their content has been used to train AI models. This should also include acknowledging the Indigenous Cultural and Intellectual Property (ICIP) rights. A licensing or remuneration framework would be appropriate to ensure Australian creators are fairly compensated for their work.
Modernise IP and copyright frameworks to provide clarity on AI-assisted and AI-generated works and support innovation while protecting creators' rights.
Modernise Australia's intellectual property and copyright frameworks to provide clarity on AI-assisted and AI-generated works and support innovation while protecting creators' rights.
Enable responsible AI research and development through balanced text and data mining provisions, greater transparency of AI training datasets, and mechanisms for attribution, licensing and remuneration.
Enable responsible AI research and development through balanced text and data mining provisions, greater transparency of AI training datasets, and appropriate mechanisms for attribution, licensing, and remuneration.
Strengthen alignment between IP protection, data sovereignty and sovereign AI capability to safeguard Australian research, creative content and strategic knowledge assets.
Strengthen the alignment between intellectual property protection, data sovereignty, and sovereign AI capability to safeguard Australian research, creative content, and strategic knowledge assets.
Support research detecting generated media beyond artefact detection and authentication of content origin and integrity.
To counter this, Australia should support new research into the detection of generated media that goes beyond artefact detection, and develop authentication methods for the origin and integrity of content to be verified.
Invest in AI-enabled defensive methods and improve workforce capability.
It should be assumed that AI will be increasingly used in both hostile and defensive settings, and appropriate investments should be made to develop AI-enabled defensive methods and improve workforce capability.
Provide monitoring and accountability for companies integrating AI into live systems, alongside research on capturing and reporting model uncertainty.
Appropriate monitoring and accountability mechanisms for companies integrating AI into live systems are needed, coupled with further research to better capture and report uncertainty within AI models.
Condition Australian market access on responsibilities to support Australian capability and capacity; restrict trade designed to increase dependence, such as vendor lock-in and social media dependence.
Access to Australian markets should necessarily require responsibilities to support Australian capability and capacity, and explicitly restrict trade designed to increase dependence (e.g. vendor lock-in, social media dependence).
Invest in research, testing and assurance frameworks to detect AI-generated content, support content authentication, test AI systems under adversarial conditions and strengthen resilience to AI-enabled cyber threats, misinformation, fraud and attacks on critical infrastructure.
Strengthen AI safety, assurance and security capabilities: Invest in research, testing, and assurance frameworks that improve the detection of AI-generated content, support content authentication, assess AI systems under adversarial conditions, and strengthen Australia's resilience to AI- enabled cyber threats, misinformation, fraud, and attacks on critical infrastructure.
Develop national AI awareness, education and workforce initiatives covering opportunities, risks, digital and critical thinking skills, workforce development, education pathways and long-term cognitive capability.
Develop national AI awareness, education, and workforce initiatives that help Australians understand the opportunities and risks of AI, strengthen digital and critical thinking skills, and address the potential impacts of AI on workforce development, education pathways, and long-term cognitive capability.
Make some Australian compute infrastructure available to Australian researchers, startups, governments, public-interest applications and domestic users on viable and dependable terms.
Some proportion of Australia's compute infrastructure should be available to Australian researchers, startups, governments, public-interest applications, and domestic users on viable and dependable terms.
Where public resources, approvals or government support facilitate major infrastructure, consider enforceable public-benefit requirements for operators and, where appropriate, users.
Where public resources, regulatory approvals, or other forms of government support facilitate major infrastructure investments, consider enforceable requirements that contribute to public benefit. Such requirements should apply not only to entities operating AI infrastructure but also, where appropriate, to those using it.
Establish a dedicated AI Skills and Workforce Capability Framework under the National AI Plan to coordinate workforce development, AI literacy, education, training, research capability and lifelong learning.
Establish a dedicated AI Skills and Workforce Capability Framework under the National AI Plan to coordinate workforce development, AI literacy, higher education, vocational training, research capability, and lifelong learning across Australia's AI ecosystem.
Link public investment in AI infrastructure to measurable capability outcomes, including researcher access, workforce development, public-interest applications, industry innovation and sovereign expertise.
Ensure that public investment in AI infrastructure is linked to measurable capability outcomes, including researcher access, workforce development, public-interest applications, industry innovation, and the growth of sovereign Australian expertise.
Support dependable Australian access to sufficient compute, including advanced accelerators, and public infrastructure that is publicly owned, governed or subject to enforceable access duties rather than solely commercial priorities.
Sovereign compute: dependable Australian access to sufficient computational resources, including advanced accelerators required for AI research, development, and deployment. • Public or publicly accessible infrastructure: capacity that is publicly owned, publicly governed, or subject to enforceable public-access obligations, rather than access being determined entirely by commercial priorities.
Develop local model capability for Australian institutions, laws, data and needs, sufficient to avoid complete dependence without reproducing every overseas frontier model; develop the workforce to operate, secure, govern, evaluate and deploy AI.
Australian model capability: the ability to develop, adapt, evaluate, and operate AI models locally, including models designed for Australian institutions, laws, data, and needs. Sovereignty does not require Australia to reproduce every frontier model developed overseas, but it does require sufficient capability to avoid complete dependence on them. • A skilled domestic workforce: the people required to develop, operate, secure, govern, evaluate, and deploy these systems.
Embed work redesign and job quality in Australia's AI workforce strategy, including human-AI task allocation, skills pathways, professional judgement, workload and autonomy.
Embed work redesign and job quality in Australia’s AI workforce strategy, including human–AI task allocation, skills pathways, professional judgement, workload and autonomy.
Consider shared or intermediary AI capability enabling smaller organisations to access trusted expertise, implementation support, infrastructure and sector-specific solutions without independently developing all these capabilities.
Australia should consider shared or intermediary models of AI capability that allow smaller organisations to access trusted expertise, implementation support, infrastructure and sector-specific solutions without requiring each organisation to develop these capabilities independently.
Develop a national AI Adoption and Value Realisation Framework measuring progression from availability and experimentation to operational adoption, effectiveness and realised value.
Develop a national AI Adoption and Value Realisation Framework that measures progression from AI availability and experimentation through operational adoption, effectiveness and realised value.
Develop national indicators of AI-enabled process improvement, decision quality, productivity, service and workforce outcomes, rather than relying primarily on tool access or reported adoption.
Develop national indicators that assess AI-enabled process improvement, decision quality, productivity, service outcomes, workforce outcomes and realised benefits, rather than relying primarily on measures of tool access or reported adoption.
Establish targeted AI adoption and capability initiatives for SMEs and regional Australia combining technology access with implementation support, trusted expertise, workforce development, data readiness and sector-specific use cases.
Establish targeted AI adoption and capability initiatives for SMEs and regional Australia that combine access to technology with implementation support, trusted expertise, workforce development, data readiness and sector-specific use cases.
Foster broad public-private partnerships between government, industry, universities and community organisations to strengthen AI capability, adoption and value realisation, aligned with broader public interests.
Foster broad public-private partnerships between government, industry, universities, and community organisations to strengthen AI capability development, adoption, and value realisation across Australian organisations while ensuring alignment with broader public interests.