AI Inquiry Atlas

An unofficial, independent guide to the public records of the Joint Select Committee on Artificial Intelligence. Not connected with the Parliament of Australia or the committee. Its makers also made a submission (no. 481; see About). Summaries and scores are written by AI and can be wrong: check the original before relying on them.

Submission 316

Commonwealth Bank of Australia

Submitter group: Companies.

An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.

Open in the interactive guide

Summary

Read the original submission for the full text.

CBA argues Australia should become a high-capability AI country through connected capability in adoption, compute, energy, skills, commercialisation and security, without competing head-on in frontier-model development. It favours existing laws with targeted AI obligations for genuine gaps. It says productivity gains and future capabilities remain uncertain, and benefits should be widely shared.

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.

  1. CBA argues Australia should set a national ambition to become a high-capability AI country on Australian terms.

    Australia should therefore set a clear national ambition to become a high-capability AI country in time to make the most of global AI transformation, on Australian terms.

    A program corrected obvious text-recognition errors in this quote for display.

  2. CBA proposes selective, connected AI capability, world-class strengths and sufficient access and resilience elsewhere, including frontier-system use and assurance without competing head-on in frontier-model development.

    Australia should build selective but connected capability across the AI value chain, with world-class strengths where it can create durable value and sufficient access, expertise and resilience elsewhere. This includes the ability to use, adapt, evaluate and secure frontier systems without competing head-on in global frontier-model development.

    A program corrected obvious text-recognition errors in this quote for display.

  3. CBA argues a predictable settlement on AI training, built on consent, control and fair payment, would give developers confidence to invest in onshore training infrastructure.

    A predictable and workable Australian settlement for AI training, built on consent, control and fair payment, would give developers confidence to invest in onshore training infrastructure while ensuring Australian creators and cultural production continue to be valued and rewarded.

    A program corrected obvious text-recognition errors in this quote for display.

  4. CBA argues large AI infrastructure projects should underwrite the additional generation and network costs they create.

    Large AI infrastructure projects should underwrite additional generation, meet the connection and network costs they create, and provide storage or flexible demand where technically and commercially viable.

    A program corrected obvious text-recognition errors in this quote for display.

  5. CBA argues the largest economic gains depend on broad workforce capability to apply AI within existing areas of expertise.

    The second capability needs to be broad and domestic. The largest economic gains will depend on managers, professionals and frontline workers being able to apply AI within the industries and institutions where they already hold deep expertise.

    A program corrected obvious text-recognition errors in this quote for display.

  6. CBA argues the AI Safety Institute should develop independent model evaluation capability and provide evidence to regulators and industry.

    The Australian AI Safety Institute should evolve its capability to independently evaluate advanced models, identify emerging risks, learn from incidents and provide evidence to regulators and industry.

    A program corrected obvious text-recognition errors in this quote for display.

Recommendations

Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.

  1. Set a clear national ambition to become a high-capability AI country in time to make the most of global AI transformation, on Australian terms.

    Australia should therefore set a clear national ambition to become a high-capability AI country in time to make the most of global AI transformation, on Australian terms.

    A program corrected obvious text-recognition errors in this quote for display.

  2. Build selective, connected AI capability: world-class strengths where durable value is possible and sufficient access, expertise and resilience elsewhere; use, adapt, evaluate and secure frontier systems without competing head-on in frontier-model development.

    Australia should build selective but connected capability across the AI value chain, with world-class strengths where it can create durable value and sufficient access, expertise and resilience elsewhere. This includes the ability to use, adapt, evaluate and secure frontier systems without competing head-on in global frontier-model development.

    A program corrected obvious text-recognition errors in this quote for display.

  3. Build globally competitive applied-AI industries where Australian expertise, lawfully usable high-quality data, sophisticated local customers and globally relevant problems intersect, creating firms, IP, exports and higher-value work, including through frontier-adjacent mining/agriculture sectors.

    Australia should also build new globally competitive applied-AI industries. We should concentrate on areas where deep Australian expertise, high-quality and lawfully usable data, sophisticated local customers and globally relevant problems intersect. The aim is to create new firms, intellectual property, exports and higher- value work that are competitive in global markets, and leverage sectors that are already mature and close to the global technology frontier, such as mining and agriculture.

    A program corrected obvious text-recognition errors in this quote for display.

  4. Build sector-based AI extension services through trusted industry, professional and regional networks to help firms identify high-value uses, improve data and workflows, train people and evaluate results.

    Australia should build sector-based AI extension services through trusted industry, professional and regional networks to help firms identify high-value uses, improve data and workflows, train people and evaluate results.

    A program corrected obvious text-recognition errors in this quote for display.

  5. Apply the same discipline to government use through GovAl and the APS AI Plan, measuring success through embedded use, productivity and better services.

    Government should apply the same discipline through GovAl and the APS A/ Plan. Success should be measured through embedded use, higher productivity and better products and services.

  6. Develop agentic-commerce policy alongside technology, protocols and industry standards, supporting transparency about agents and representation, participant risk management, consumer/merchant safety and consent, privacy, choice and interoperability.

    That means transparency about when agents are involved and who they represent; safety supported by participants across the value chain working together to manage risk; privacy and data use grounded in consumer and merchant safety and consent; genuine choice over the services people and businesses use; and interoperability that allows them to connect across the agentic commerce ecosystem. These foundations should develop alongside technology, protocols, policy and industry standards so that trust supports the safe and sustainable growth of agentic commerce in the Australian market.

  7. Build globally significant digital infrastructure combining cloud/inference for adoption, research/model-adaptation compute for domestic capability, and training-grade capacity where durable commercial demand and national benefit exist; leave energy, grid, skills, research/compute access and supply-chain benefits.

    Australia should build globally significant digital infrastructure that positions us as a trusted regional base for AI compute and intelligence production. This means a balanced mix of cloud and inference capacity for broad adoption, research and model-adaptation compute for domestic capability, and training-grade capacity where there is durable commercial demand and national benefit. Major investment should leave more behind than physical infrastructure: energy and grid capability, skills, research access, compute access, resilient supply chains and stronger links to Australian firms.

    A program corrected obvious text-recognition errors in this quote for display.

  8. Distinguish inference, research and training needs in a compute roadmap; provide public-interest research capacity, secure critical-function capacity, affordable access for promising firms, diversified supply and portability; scale training capacity where demand and national benefit justify.

    Inference will support adoption across the economy, while research compute can widen access for universities and emerging firms. Training-grade infrastructure can enable Australian organisations to train and adapt advanced models using local data and expertise, particularly as powerful open-weight models become more widely available. A national compute roadmap should distinguish these needs and provide for public-interest research capacity, secure capacity for critical functions, affordable 5 Commonwealth Bank of Australia access for promising firms, and diversified supply and portability. Training-grade capacity should scale where demand and national benefit justify it.

  9. Provide clear, practical, enforceable rules for advanced AI, including a predictable and workable Australian settlement for AI training built on consent, control and fair payment.

    The rules governing advanced AI should also be clear, practical and enforceable. A predictable and workable Australian settlement for AI training, built on consent, control and fair payment, would give developers confidence to invest in onshore training infrastructure while ensuring Australian creators and cultural production continue to be valued and rewarded.

    A program corrected obvious text-recognition errors in this quote for display.

  10. Require global technology companies to comply with Australian law and make an appropriate economic contribution, and large AI training projects to underwrite additional energy and use closed-loop cooling.

    Global technology companies should comply with Australian law and make an appropriate economic contribution here, while large AI training projects should underwrite the additional energy they require, and operate with closed-loop cooling so as not to draw on water resources.

    A program corrected obvious text-recognition errors in this quote for display.

  11. Where major AI infrastructure receives public facilitation, accelerated approvals or constrained resources, require clear, transparent, verifiable additional Australian benefit, potentially energy/storage, regional contributions, skills/apprenticeships, engineering/research, priority researcher/emerging-firm compute access, supplier links or cyber resilience.

    Where major AI infrastructure projects receive public facilitation, accelerated approvals or access to constrained resources, the additional benefit to Australia should be clear, transparent and verifiable. This could include additional energy and storage, contributions to regional economies, local skills and apprenticeships, Australian engineering and research, priority access to compute for researchers and emerging firms, stronger links to Australian suppliers, or cyber resilience.

    A program corrected obvious text-recognition errors in this quote for display.

  12. Make programs for new AI infrastructure capability accessible and useful to local businesses, workers and communities.

    Programs should also make new AI capability accessible and useful to local businesses, workers and communities.

    A program corrected obvious text-recognition errors in this quote for display.

  13. Build generation, storage, firming and transmission ahead of demand, with timely connections and coordinated planning for major new loads.

    That will require generation, storage, firming and transmission to be built ahead of demand, supported by timely connections and coordinated planning for major new loads.

  14. Have large AI infrastructure projects underwrite additional generation, meet the connection and network costs they create, and provide storage or flexible demand where technically and commercially viable.

    Large AI infrastructure projects should underwrite additional generation, meet the connection and network costs they create, and provide storage or flexible demand where technically and commercially viable.

    A program corrected obvious text-recognition errors in this quote for display.

  15. Build deeper domestic pathways through universities, research institutions and industry.

    Building deeper domestic pathways through universities, research institutions and industry will be equally important.

  16. Account for wider workforce pressures in planning AI infrastructure, aligning skills, migration and training settings accordingly.

    Planning for AI infrastructure should therefore account for these wider workforce pressures and align skills, migration and training settings accordingly.

    A program corrected obvious text-recognition errors in this quote for display.

  17. Have Jobs and Skills Australia maintain a standing view of how AI is changing tasks, occupations and skills demand.

    Nationally, Jobs and Skills Australia is well placed to maintain a standing view of how AI is changing tasks, occupations and skills demand, allowing governments, education providers and employers to respond.

    A program corrected obvious text-recognition errors in this quote for display.

  18. Build a stronger pathway from research and start-ups to first customers, growth capital, global scale and intellectual property.

    Australia should build a stronger pathway from research and start-ups to first customers, growth capital, global scale and intellectual property.

  19. Government and large businesses should make it easier for Australian firms to move from trials to commercial contracts.

    Government and large businesses can play an important role by making it easier for Australian firms to move from trials to commercial contracts.

  20. Keep Australian intellectual property and creators' rights protected as global technology partnerships grow.

    Australian intellectual property and creators’ rights should remain protected as these partnerships grow.

  21. Ensure Australian institutions can access, evaluate and safely deploy advanced AI systems at the pace required to keep up.

    Australia should ensure its institutions can access, evaluate and safely deploy advanced systems at the pace required to keep up.

  22. Evolve the Australian AI Safety Institute's capability to independently evaluate advanced models, identify emerging risks, learn from incidents and provide evidence to regulators and industry.

    The Australian AI Safety Institute should evolve its capability to independently evaluate advanced models, identify emerging risks, learn from incidents and provide evidence to regulators and industry.

    A program corrected obvious text-recognition errors in this quote for display.

  23. Ground AI regulation in existing sectoral and technology-neutral laws, with targeted AI-specific obligations where genuine gaps emerge.

    This capability can support a regulatory approach grounded in existing sectoral and technology-neutral laws, with targeted AI-specific obligations where genuine gaps emerge.

    A program corrected obvious text-recognition errors in this quote for display.

  24. For consequential uses of AI, keep clear accountability, transparency and effective human review as central safeguards.

    For consequential uses of AI, clear accountability, transparency and effective human review should remain central safeguards.

    A program corrected obvious text-recognition errors in this quote for display.

Read the original

In some quotes a program corrected obvious text-recognition errors for display.

Open in the interactive guide