Submission 518
Australian Alliance for Indigenous Genomics (ALIGN)
Submitter group: Research institutes.
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 systems using Aboriginal and Torres Strait Islander data should be governed by Indigenous data sovereignty principles, with Indigenous governance legislated as a requirement. Also addresses health AI with human oversight, sovereign compute, workforce pathways, informed consent, privacy and copyright protection for Indigenous cultural knowledge.
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 Indigenous governance should be a legislated requirement for AI development and use, which it says would place Australia in a pre-eminent position globally.
Incorporating Indigenous governance as a legislated requirement for the development and use of AI technologies will set Australia apart and in a pre-eminent position globally.
Says Australian AI Standards should have Indigenous governance bodies control permitted uses and the release of weights, outputs and evaluation data for models trained on Aboriginal and Torres Strait Islander data.
Australian AI Standards should recommend the development of models trained on Aboriginal and Torres Strait Islander data, and stored and evaluated within the Australian jurisdiction, with Indigenous governance bodies controlling permitted uses and the release of weights, outputs, and evaluation data.
Recommends human-in-the-loop accountability for all high-risk AI applications, particularly health research and service delivery.
We therefore strongly recommend that a “human in the loop” accountability model be built within all high-risk AI applications and specifically within health research and service delivery environments where the risks for inaccurate or insufficient data and synthesis may adversely impact the health or wellbeing outcomes of patients.
States that AI technologies that use sensitive healthcare data to train and refine models should require explicit consent and provide additional protections addressing Indigenous self-determination, data sovereignty and governance, and cultural knowledges.
AI technologies that use sensitive healthcare data to train and refine models should require explicit consent from a consumer or patient and specifically recognise and provide additional protections or regulations that address Indigenous self-determination, data sovereignty and governance, and cultural knowledges and intellectual property.
Recommends legislation requiring training-data documentation, licences where required and affordable challenges to unauthorised use, alongside Indigenous-led protections for artists and Communities.
We strongly advocate for the development of Australian legislation that mandates for AI developers to document their training-data sources and the legal basis for use, obtain licences where required, and provide affordable routes for creators to challenge unauthorised use. Indigenous-led legislation should address AI training and generated content, protect individual artists' rights and Communities' cultural authority.
Argues that Australia needs the capacity to develop, test and run AI models under Australian control, with the people and infrastructure to maintain them.
Australia also needs the capacity to develop, test and run AI models under Australian control, with the people and infrastructure to maintain them.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Develop and apply specific AI guidelines or regulations for AI systems using Indigenous data, and culturally informed data sovereignty and governance frameworks guiding use of Indigenous data, consent processes, storage and future research and applications.
Specific AI guidelines or regulations be developed and applied to AI systems utilising Indigenous data and information and for the development of culturally informed data sovereignty and governance frameworks that guide the use of Indigenous data, consent processes, storage and future research and applications.
Develop best practice approaches that protect Indigenous self-determination, values, languages and free, prior and informed consent, such as those under Indigenous Data Sovereignty and the FAIR and CARE principles.
The development of best practice approaches that protect Indigenous self-determination, values, languages, and free, prior and informed consent, such as those provided under Indigenous Data Sovereignty (IDSov) and the FAIR and CARE principles;
Build all AI systems that rely on Australian datasets with the data sovereignty and governance frameworks pioneered and co-developed by Aboriginal and Torres Strait Islander peoples.
ALIGN strongly advocates that all AI systems that rely on Australian datasets be built with the data sovereignty and governance frameworks pioneered and co-developed by Aboriginal and Torres Strait Islander peoples.
Incorporate Indigenous governance as a legislated requirement for the development and use of AI technologies.
Indigenous governance be incorporated as a legislated requirement for the development and use of AI technologies,
Build a "human in the loop" accountability model into all high-risk AI applications, specifically in health research and service delivery where inaccurate data could harm patient outcomes.
A “human in the loop” accountability model be built within all high-risk AI applications and specifically within health research and service delivery environments where there is a risk for inaccurate or insufficient data to adversely impact the health or wellbeing outcomes of patients.
Provide regulatory and policy guidelines ensuring that clinicians or researchers who review and approve AI results have ethically and legally performed their duties and do not increase patient risk through a lack of oversight or understanding of the AI tools.
Regulatory and policy guidelines, such as the Australian Commission on Safety and Quality in Health Care,7 are needed to ensure that when clinicians or researchers are required to review and approve AI results (clinical notes, examination summaries, diagnostic test results), they have ethically and legally performed their duties and do not increase the potential for physical or emotional patient risk or harm through a lack of oversight or understanding of the AI tools they utilise in patient care
Endorse open access AI principles while defending community-specific guidelines and cultural safety protections such as Indigenous Data Sovereignty and Indigenous Cultural and Intellectual Property.
Government endorsement of open access AI principles while defending community-specific guidelines and cultural safety protections such as Indigenous Data Sovereignty (IDSov) and Indigenous Cultural Intellectual Property (ICIP), would significantly support the Australian health and medical research sectors and enable AI systems to be developed more equitability and transparently.
Revise existing copyright and intellectual property laws to protect Indigenous knowledges and cultural intellectual property and reduce misuse and exploitation in AI.
Existing copyright and intellectual property laws be revised to protect Indigenous knowledges and cultural intellectual property to reduce its misuse and exploitation within AI environments and technologies.
Consider a disruptive approach to existing regulatory frameworks, such as co-developing and embedding Indigenous governance and sovereignty models within AI systems.
A disruptive approach to existing antiquated regulatory frameworks be considered – for example by co-developing and embedding Indigenous governance sovereignty and governance models within AI systems
Develop an Australian AI health workforce pathway with Genomics Australia, states and territories and Aboriginal and Torres Strait Islander partners, with Indigenous-led and regional placements, doctoral and conversion funding, retraining support and entry-level monitoring.
An Australian AI health workforce pathway be developed with Genomics Australia, the states and territories and key Aboriginal and Torres Strait Islander partners - supporting Indigenous-led and regional placements - alongside doctoral and conversion funding. Workforce policy should also support consultation and retraining for people whose roles change and monitor entry-level opportunities as AI use expands.
Require Australian storage and evaluation of Indigenous-data models; give Indigenous bodies authority over uses and releases of weights, outputs and evaluation data, with the AI Safety Institute developing safeguard tests alongside those bodies.
Australian AI Standards should embed the requirement for models to be trained on Aboriginal and Torres Strait Islander data, stored and evaluated within the Australian jurisdiction and with Indigenous governance bodies controlling permitted uses and the release of weights, outputs and evaluation data. The AI Safety Institute should work with those bodies to develop methods for testing these safeguards. Community authority must determine how the data and models are used; their location alone is not sufficient.
Extend Indigenous data governance to model training, evaluation and release as well as access to underlying data.
Indigenous data governance should cover the training, evaluation, and release of models, as well as access to the underlying data.
Provide sustained national funding for Indigenous governed AI research, data, computing and engineering, and guarantee that all data remain in Australia.
Sustained national funding for Indigenous governed AI research, data, computing and engineering, and guarantees that all data remain in Australia.
Prioritise health and genomics in the National AI Plan, measuring independently validated performance across intended Australian populations; require Indigenous governance and Community priorities for research involving Aboriginal and Torres Strait Islander people and data.
Prioritising health and genomics within the National AI Plan, along with progress measured by independent validation on Australian cohorts and performance across the populations intended to benefit. Research involving Aboriginal and Torres Strait Islander people and data must be directed by Indigenous governance and Community priorities.
Consult Aboriginal and Torres Strait Islander Communities, Traditional Owner Groups, Prescribed Body Corporations and health services on commercial data centres’ priorities, concerns, opportunities and impacts.
we strongly recommend that extensive consultations be undertaken with Aboriginal and Torres Strait Islander Communities and organisations (including Traditional Owner Groups, Prescribed Body Corporations, Aboriginal and Torres Strait Islander health services) to assess their priorities, concerns, opportunities and impacts that large scale commercial data centres may present.
Invest in shared Australian GPUs, secure data and engineering; offer small businesses predictable-price managed services without buying hardware or hiring specialists; fund operations, maintenance and reserves, migration, portability and quality-preserving switching checks.
We recommend substantial investment in Australian infrastructure for running and serving AI models, including shared GPU capacity, secure data systems, and local engineering teams. Small businesses should be able to use this infrastructure through affordable managed services with predictable pricing, without having to buy hardware or employ their own specialist teams. Funding must cover ongoing operations, maintenance, and reserve capacity as well as equipment. Support should include migration assistance, portable data and workflows, and regular checks that businesses can switch models while maintaining the quality their applications require.
Fund affordable access, practical training and ongoing technical assistance for Aboriginal and Torres Strait Islander businesses and community-controlled organisations, involving those organisations in service design.
Funding should support affordable access, practical training and ongoing technical assistance for Aboriginal and Torres Strait Islander businesses and 7Australian Commission on Safety and Quality in Healthcare, AI Clinical Use Guide – Guidance for clinicians, AI Clinical Use Guide – Guidance for clinicians 8 Error by AI scribe during medical appointment leaves patient devastated - ABC News 9 Rajpurkar, P. et al. (2021). Underdiagnosis bias of artificial intelligence algorithms applied to chest radiographs in under-served patient populations. Nat. Med. 27, 2176–2182. https://doi.org/10.1038/s41591-021-01595-0 community-controlled organisations, with those organisations involved in designing the services.
For regional and remote users, address connectivity costs and provide services that work with limited bandwidth.
For regional and remote users, this should include addressing connectivity costs and providing services that work with limited bandwidth.
Have relevant Indigenous governance bodies decide who may access community data and whether it can be reused to train or improve models, with those conditions preserved when a business switches providers.
Relevant Indigenous governance bodies should decide who may access community data and whether it can be reused to train or improve models, with those conditions preserved when a business switches providers.
Define, address and embed informed consent in AI use, particularly in high-risk environments; require explicit consent for use of sensitive healthcare data in model training, with additional protections addressing Indigenous self-determination, data sovereignty, governance, cultural knowledges and intellectual property.
We strongly urge the Australian government to define, address, and embed the tenants of informed consent within the use and application of AI, particularly in high-risk environments. AI technologies that use sensitive healthcare data to train and refine models should require explicit consent from a consumer or patient and specifically recognise and provide additional protections or regulations that address Indigenous self-determination, data sovereignty and governance, and cultural knowledges and intellectual property.
Conduct a robust review and analysis of whether informed consent in research and in receiving health care also covers the use of AI systems and techniques in delivering those services.
A robust review and analysis for examining whether informed consent in research and for receiving health care also includes consent for the use of AI systems and techniques within the delivery of those services.
Develop a federal privacy framework establishing healthcare data and AI guardrails and enabling prosecution of individuals or companies failing to secure personal or potentially identifiable information.
The Australian government needs to develop a federal framework for data privacy that would help establish guardrails for healthcare data and AI usage and enable the prosecution of individuals or companies that do not adequately secure personal or potentially identifiable information.
Legislate that AI developers document training-data sources and the legal basis for use, obtain licences where required, and provide affordable routes for creators to challenge unauthorised use.
We strongly advocate for the development of Australian legislation that mandates for AI developers to document their training-data sources and the legal basis for use, obtain licences where required, and provide affordable routes for creators to challenge unauthorised use.
Enact Indigenous-led legislation addressing AI training and generated content, requiring free, prior and informed consent for protected cultural material with attribution and fair payment or benefit-sharing, letting Communities refuse use of sacred or restricted material, with remedies against false Indigenous attribution and funding for community-controlled legal and licensing support.
Indigenous-led legislation should address AI training and generated content, protect individual artists' rights and Communities' cultural authority. Use of protected cultural material should require free, prior, and informed consent, with agreed attribution and fair payment or benefit-sharing. Communities should be able to refuse use of sacred or restricted material. These protections should include remedies against false Indigenous attribution and funding for community-controlled legal and licensing support.
Undertake a considered evaluation of Indigenous knowledge systems, in partnership with Indigenous peoples and with Indigenous leadership, in respectful consultation.
A considered evaluation of these unique knowledge systems in partnership with Indigenous peoples must be undertaken in respectful consultation and with Indigenous leadership.
Consider opportunities to formally integrate established open-access standards and best practices, such as UNESCO’s Recommendation on Open Science, into future regulatory and policy options.
For any future regulatory and policy options, we recommend that the government consider opportunities to formally integrate well-established open access standards and best practices, such as the UNESCO Recommendation on Open Science2, as part of this consultation.
Invest in research compute infrastructure (such as NCRIS and BioCommons) so researchers can build novel systems for the Australian healthcare system.
Australia needs to invest in research compute infrastructure (NCRIS, BioCommons etc) to ensure researchers can safely build novel systems for the Australian healthcare system.
Invest significantly in specialised AI training programs, especially for Aboriginal and Torres Strait Islander students and people living in remote regions.
Significant investment needs to be made in specialised training programs, especially for Aboriginal and Torres Strait Islander students and people living in Australia’s remote regions.
Align with the Closing the Gap national strategy and confirm the Australian Government's commitment to UNDRIP in AI governance.
Alignment with the Closing the Gap national strategy to improve the health and well-being of Aboriginal and Torres Strait Islander peoples and confirming the Australian Government's commitment to the UNDRIP is essential.
Support a whole-of-economy regulatory approach to minimise conflicting rules and improve collaboration, enforcement, interoperability and oversight, alongside Indigenous-led data sovereignty protections.
Without appropriate Indigenous-led and developed data sovereignty protection mechanisms and models, mandatory mainstream regulations in isolation will not be adequate nor provide the level of appropriate stewardship and governance that is required. Supporting a whole-of-economy approach to AI regulation to minimise conflicting rules and enhance national and international AI collaboration and innovation will also deliver greater consistency in legal enforcement, interoperability, and oversight.
Guarantee equity and justice through reforms and systematic evaluation.
Systematic evaluation is essential and reforms that guarantee equity and justice for all Australians is a must.
Invest in paid training combining health or genomics with software and data engineering.
Australia’s investment into paid training that combines health or genomics with software and data engineering is 17 Rye, C et al 2025, “Rakeora Genomics Platform: a path finder for genomic medicine research in Aotearoa New Zealand”, Rakeiora Genomics Platform: a pathfinder for genomic medicine research in Aotearoa New Zealand - Rye - 2025 - Journal of the Royal Society of New Zealand - Wiley Online Library 18 NCIG Statute 2021 https://www.legislation.gov.au/F2021L00183/asmade/text 19 Global Indigenous Data Alliance. CARE Principles for Indigenous Data Governance. https://www.gida-global.org/care essential for building AI capacity and capability.