Submission 283
Australian Information Industry Association (AIIA)
Submitter group: Industry bodies.
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.
The Australian Information Industry Association argues for an adoption-focused national AI agenda: SME adoption pathways, a broad view of sovereign capability, reusable government procurement, a gap-based nationally coherent regulatory approach, voluntary copyright licensing, and attention to safeguards, cyber security, compute and data-centre infrastructure.
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 AI presents a significant national opportunity for productivity, services, investment and jobs if managed well.
Artificial intelligence presents a significant national opportunity for Australia. Managed well, it can improve productivity, strengthen public services, attract investment, support sovereign capability, create and augment jobs and enable Australian businesses to compete in global markets.
Argues that SME adoption programs should be practical and use-case led rather than scaled-down enterprise programs.
Government should avoid designing SME programs as smaller versions of large-enterprise transformation initiatives. Support should be practical and use-case led.
Argues that sovereign AI capability should not be equated with technological self-sufficiency, as Australia will continue to rely on global technology companies, cloud providers, semiconductor supply chains and international research.
Sovereign AI capability should not be equated with technological self-sufficiency. Australia will continue to rely on global technology companies, cloud providers, semiconductor supply chains and international research.
Argues that, with governance foundations already in place, the Commonwealth's next phase should focus on implementation quality and measurable outcomes.
We are already beyond recommending basic governance structures that already exist. The next phase should focus on implementation quality and measurable outcomes
Recommends a gap-based approach to regulatory reform that tests existing law first.
AIIA supports a gap-based approach. Government should identify the harmful outcome or accountability failure, test an existing law, and only then determine whether additional rules are required.
Recommends market-based voluntary licensing for AI training inputs, proportionate transparency, and guidance protecting sufficiently human-authored AI-assisted works.
AIIA supports market-based voluntary licensing for AI training inputs, proportionate transparency and clear guidance protecting sufficiently human-authored AI-assisted works.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Set measurable national AI adoption and productivity objectives, with transparent measures of adoption, productivity, service quality, investment, commercialisation, workforce capability and public trust accompanying National AI Plan implementation.
Set measurable national AI adoption and productivity objectives. Implementation of the National AI Plan should be accompanied by transparent measures of adoption, productivity, service quality, investment, commercialisation, workforce capability and public trust.
Create SME, family-business and regional AI adoption pathways covering trusted advice, affordable use cases, skills, demonstrations, secure technology, regulator-endorsed compliance pathways and support from pilot to scale.
Create practical and legally certain adoption pathways for SMEs, family businesses and regional Australia. Support should include trusted advice, affordable use cases, skills, demonstration projects, secure technology, regulator-endorsed compliance pathways and support from pilot to scale.
Build sovereign AI capability across data, compute, research, talent and domain expertise, strengthening strategic choice and resilience without requiring every layer of the global AI stack to be built domestically.
Build sovereign AI capability across data, compute, research, talent and domain expertise. Sovereignty should strengthen strategic choice and resilience without requiring Australia to build every layer of the global AI stack domestically.
Strengthen research commercialisation and use government as a lead customer, with procurement, challenge programs and testbeds moving Australian AI and robotics firms from pilots to reference customers and export.
Strengthen research commercialisation and use government as a lead customer. Procurement, challenge programs and testbeds should help Australian AI and robotics firms move from research and pilots to reference customers, scale and export markets.
Develop reskilling, change management and incentives maintaining workforce investment alongside AI deployment.
Build a national AI skills and workforce-transition pipeline. Foundational literacy, practitioner capability and specialist skills should be developed alongside reskilling, change management and incentives that maintain workforce investment during AI deployment.
Have DTA and Finance establish shared architectures, reusable components, security/privacy/assurance materials, standard contract clauses and modular procurement; expect appropriate agency reuse with accountability for local context and risks.
The Digital Transformation Agency and Department of Finance should establish shared reference architectures, reusable components, common security, privacy and assurance materials, standard contract clauses and modular procurement pathways. Agencies should be expected to reuse these foundations where appropriate, while remaining accountable for their own operating context and risks.
Require open data formats, documented standards-based APIs/protocols, interoperable metadata, recognised identity/access standards and technically appropriate portability of data, workflows and relevant model artefacts in Commonwealth AI procurement.
Open standards should be an explicit part of Commonwealth AI architecture and procurement. Government systems should support open data formats, documented and standards-based APIs and protocols, interoperable metadata, recognised identity and access standards, and the practical portability of data, workflows, and relevant model artefacts where technically appropriate.
Maintain a nationally coherent, technology-neutral, risk-based regulatory architecture: existing laws as the baseline, with targeted reform for demonstrated gaps that does not duplicate obligations or create conflicting state and federal regimes.
Maintain a nationally coherent, technology-neutral and risk-based regulatory architecture. Existing laws should remain the baseline; targeted reform should address demonstrated gaps without duplicating obligations or creating conflicting state and federal regimes.
Use market-based voluntary licensing for AI training inputs.
AIIA supports market-based voluntary licensing for AI training inputs
Strengthen prevention and enforcement against AI-enabled harms, with online safety, consumer protection and other regimes addressing scams, deepfakes and harms to children and vulnerable people, supported by safety-by-design, provenance and redress.
Strengthen prevention and enforcement against AI-enabled harms. Online safety, consumer protection and other regimes should address scams, deepfakes and harms to children and vulnerable Australians, supported by safety-by-design, provenance and effective redress.
Use the AI Safety Institute for technical evaluation and evidence, complementing ASD, intelligence agencies and existing regulators.
The Australian AI Safety Institute should provide technical evaluation and evidence to regulators and agencies, complementing ASD, the intelligence community and existing regulators.
Require Commonwealth agencies and covered critical-infrastructure entities to identify unsupported critical systems and maintain time-bound replacement, upgrade or isolation plans through existing Commonwealth cyber-security and Security of Critical Infrastructure frameworks.
The Committee should recommend a risk-based national program requiring Commonwealth agencies and covered critical-infrastructure entities to identify unsupported systems that support critical functions and maintain time-bound plans to replace, upgrade or isolate them. This should operate through existing Commonwealth cyber-security and Security of Critical Infrastructure frameworks rather than creating a separate AI-specific regime.
Establish a competitive, multi-provider National Innovation Compute Reserve for Australian researchers, start-ups, scale-ups, SMEs and public agencies, potentially combining capacity, allocations, credits, testing and co-investment with transparent eligibility, security and commercialisation criteria.
To convert infrastructure investment into domestic capability, the Commonwealth should establish a National Innovation Compute Reserve through competitive, multi-provider arrangements. The reserve could combine guaranteed capacity, research allocations, cloud credits, shared testing environments and co-investment, with transparent eligibility, security and commercialisation criteria for Australian researchers, start-ups, scale-ups, SMEs and public agencies.
Track SME participation in the National AI Plan’s public measurement framework; measure success by outcomes rather than numbers of pilots or tools.
AIIA recommends that implementation of the National AI Plan be supported by a public measurement framework that tracks productivity outcomes, service improvements, business investment, workforce capability, commercialisation, SME participation and trust. Success should be measured by outcomes rather than the number of pilots or tools deployed.
Use existing business-support channels where possible for diagnostic advice, short practical training, cyber/governance checklists, demonstrations and holistic support; evaluate deployment and productivity outcomes, not participation alone.
AIIA recommends a coordinated adoption pathway using existing business-support channels wherever possible. It should include trusted diagnostic advice, short practical training, cyber and governance checklists, access to demonstration environments and holistic support services. Programs should be evaluated against deployment and productivity outcomes rather than participation numbers alone.
Work with OAIC and sector regulators on vendor-neutral architectures, pre-assessed deployment patterns and standard privacy, security and contractual controls for common lower-risk AI uses.
Government should work with the OAIC and relevant sector regulators to develop vendor- neutral reference architectures, pre-assessed deployment patterns and standard privacy, security and contractual controls for common lower-risk AI uses.
Offer compliant SMEs a defined pathway, potentially limited civil-penalty safe harbour or rebuttable compliance presumption for inadvertent breaches within approved patterns, conditional on good-faith controls and prompt problem correction.
SMEs that implement the required controls in good faith and promptly address identified problems should have a clearly defined compliance pathway. This could include a limited safe harbour from civil regulatory penalties, or a rebuttable presumption of compliance, for inadvertent breaches occurring within the approved deployment pattern.
Exclude deliberate, reckless or repeated non-compliance from SME protection; retain breach notification/remediation obligations and individuals’ correction and redress rights.
Any protection should be carefully defined. It should not cover deliberate, reckless or repeated non-compliance, remove obligations to notify and remediate breaches, or prevent affected individuals from seeking correction or redress.
Build regional adoption on existing universities, TAFEs, industry clusters, chambers and digital hubs rather than a separate bureaucracy; align demonstrations with local strengths and enable replication.
Regional adoption initiatives should build on existing universities, TAFEs, industry clusters, chambers and digital hubs rather than create a separate national bureaucracy. Demonstration projects should be tied to local economic strengths and should include a pathway for successful solutions to be replicated in other regions.
Promote AI accessibility and participation for disabled people, First Nations communities and digitally excluded Australians through inclusive design, accessible interfaces, engagement and safeguards against bias and unfair outcomes.
AI policy should also promote accessibility and participation for people with disability, First Nations communities and Australians experiencing digital exclusion. Inclusive design, accessible interfaces, community engagement and safeguards against bias and unfair outcomes are important both as matters of equity and because they expand the number of Australians able to benefit from AI-enabled services.
Keep high-impact decisions affecting individuals subject to appropriate human oversight, transparent governance and accessible review, complaint and redress pathways.
High-impact decisions affecting individuals should remain subject to appropriate human oversight, transparent governance and accessible pathways for review, complaint and redress.
Use challenge-based procurement, sandboxes, testbeds, outcome-based contracts and targeted co-investment to bridge successful pilots and commercially sustainable products.
Government can play a catalytic role through challenge-based procurement, sandboxes and testbeds, outcome-based contracts and targeted co-investment. The objective should be to reduce the gap between a technically successful pilot and a commercially sustainable product.
Coordinate implementation of the National AI Plan and the National Robotics Strategy more explicitly, particularly for industries where AI-enabled physical systems can address labour shortages, safety risks and remote operating environments.
AIIA recommends that implementation of the National AI Plan and National Robotics Strategy be coordinated more explicitly, particularly for industries where AI-enabled physical systems can address labour shortages, safety risks and remote operating environments.
Distinguish workloads by sensitivity and consequence, enable appropriate trusted cross-border arrangements rather than blanket localisation, and strengthen assurance for sensitive government, health, critical-infrastructure and national-security workloads.
Government should respond by improving clarity and choice rather than defaulting to blanket localisation. Policy should distinguish workloads by sensitivity and consequence, enable trusted cross-border arrangements where appropriate, and support stronger assurance for sensitive government, health, critical infrastructure and national-security workloads.
Invest in data quality, safe-access environments, synthetic data, privacy-enhancing technologies and clear authority for responsible data sharing, without weakening privacy, copyright, security or confidentiality protections.
Better use of Australian data should not mean weakening privacy, copyright, security or confidentiality protections. Instead, government should invest in data quality, safe-access environments, synthetic data, privacy-enhancing technologies and clear authority for responsible data sharing.
Assess government technology partnerships by retained Australian capability; welcome foreign investment while considering vendor trust, ownership/control, concentration and supply-chain integrity for sensitive workloads.
Government partnerships should be assessed by the capability they leave behind in Australia. Foreign investment should continue to be welcomed, but arrangements involving sensitive workloads should also consider vendor trust, ownership and control, concentration risk and supply-chain integrity.
Develop foundational safe/critical/productive AI literacy; practitioner training for staff selecting, implementing or supervising AI; and specialist engineering, data, cyber, robotics, evaluation, governance and assurance pathways.
AIIA recommends a national skills pipeline spanning foundational, practitioner and specialist capability. Foundational AI literacy should equip workers to use AI tools safely, critically and productively. Practitioner training should support managers, professionals and frontline staff who select, implement or supervise AI. Specialist pathways should deepen capability in AI engineering, data science, cyber security, robotics, model evaluation, governance and assurance.
Consider a time-limited AI Skills Investment Incentive, such as an enhanced tax deduction or matched co-investment for accredited AI, cyber-security and governance training, subject to Treasury modelling and integrity safeguards.
The Commonwealth should consider a time-limited AI Skills Investment Incentive, such as an enhanced tax deduction or matched co-investment for accredited AI, cyber-security and governance training. The design should be subject to Treasury modelling and appropriate integrity safeguards.
Require major Commonwealth AI grants and procurements to include proportionate plans demonstrating how employees will be prepared to implement and govern AI, avoiding rigid requirements penalising legitimate training changes.
Major Commonwealth AI grants and procurements should also require a proportionate workforce capability plan demonstrating how employees will be prepared to implement and govern the technology. The objective should be to ensure that workforce capability grows alongside technology investment, without imposing rigid requirements that penalise businesses for legitimate changes in how training is structured or funded.
Build school AI literacy, critical thinking and responsible use without making commercial tools permanent curriculum.
School education should build AI literacy, critical thinking and responsible use without treating commercial tools as permanent curriculum.
Improve labour-market monitoring and support reskilling where displacement risks are concentrated.
Government should improve labour-market monitoring and support reskilling where displacement risks are concentrated.
Train human–AI collaboration skills: evaluation, verification, bias awareness, escalation, safe delegation and automation limits; give supervising workers authority and confidence to intervene.
The emergence of agentic AI and physical AI makes human-AI collaboration especially important. Training should include critical evaluation, verification, bias awareness, escalation, safe delegation to automated systems and the limits of automation. Workers supervising AI need authority and confidence to intervene when systems behave unexpectedly.
Require each significant Commonwealth AI pilot to identify the problem, expected public value, evidence needed to reach production, integration and workforce needs, funding and retirement criteria.
Each significant pilot should identify the problem, expected public value, evidence required to proceed to production, integration and workforce requirements, ongoing funding and criteria for retirement if it does not deliver.
Favour modular, interoperable procurement enabling Australian SME and scale-up participation; give challenge programs and testbeds genuine routes to procurement and production.
Procurement should favour modular and interoperable solutions over monolithic contracts, allowing Australian SMEs and scale-ups to supply components and establish government reference customers. Challenge programs and testbeds should include a genuine pathway to procurement and production rather than treating the proof of concept as the final outcome.
Assess lock-in, concentration, migration/exit costs, portability, justified multi-cloud/hybrid deployment, licensing, resilience, security and lifecycle value; include export/deletion, transition assistance, continuity and exit in contracts.
Procurement assessments should consider vendor lock-in and concentration risk, migration and exit costs, data and workload portability, multi-cloud or hybrid deployment where justified, licensing, resilience, security and total lifecycle value. Contracts should include practical arrangements for data export and deletion, transition assistance, service continuity and exit.
Apply accountability, impact assessment, auditability, human oversight, monitoring and review to high-impact government AI; explain material influence on services or decisions and responsibility.
Government has an elevated responsibility when AI materially affects individuals. High-impact uses should be subject to clear accountability, impact assessment, auditability, human oversight, monitoring and review mechanisms. Transparency should be meaningful. Citizens should be able to understand when AI materially influences a service or decision, what role it plays and where responsibility sits.
Publish material AI use cases and outcomes while protecting personal information, cyber security, procurement confidentiality and legitimate IP.
AIIA supports public reporting that focuses on material use cases and outcomes. Government should publish enough information to build confidence while protecting personal information, cyber security, procurement confidentiality and legitimate intellectual property.
Identify harmful outcomes or accountability failures and test existing law before adding rules; impose stronger expectations on higher-risk applications without equivalent burdens on lower-risk uses.
AIIA supports a gap-based approach. Government should identify the harmful outcome or accountability failure, test an existing law, and only then determine whether additional rules are required. Possible responses may include clarifying existing duties, improving regulator powers or technical capability, or introducing targeted new obligations for defined high-impact uses. Higher-risk applications should face stronger expectations. Lower-risk uses should not be subjected to the same compliance burden merely because they use the same underlying technology. This distinction is important for SMEs and for routine productivity tools.
Have the Office of AI coordinate portfolios while preserving existing regulators’ responsibilities; clarify allocation, use common definitions/interoperable standards and enable technical-evidence and emerging-risk intelligence sharing.
The Office of AI should play a coordinating role across portfolios, but the framework should preserve the expertise and responsibilities of existing regulators. The objective should be clear allocation of responsibility, common definitions where appropriate, interoperable standards and mechanisms for regulators to share technical evidence and emerging-risk intelligence.
Support internationally interoperable risk-management, impact-assessment, security and assurance approaches, avoiding unnecessarily disconnected Australian compliance processes.
Government should support internationally interoperable approaches to risk management, impact assessment, security and assurance so Australian organisations can meet domestic obligations without creating Australia-specific compliance processes that are unnecessarily disconnected from global practice.
Encourage interoperable rights information, standard contracts, licensing marketplaces and rights-preference/access mechanisms; address genuine market failures without pre-empting developing commercial models absent evidence they cannot function.
Government should encourage interoperable rights information, standard contractual approaches, licensing marketplaces and technical mechanisms that allow rightsholders to express preferences and developers to obtain lawful access efficiently. Policy should be alert to genuine market failures, but intervention should not pre-empt developing commercial models without evidence that they cannot function.
Issue formal guidance confirming Australian copyright law protects works created by humans using AI tools where there is sufficient human authorship.
AIIA has also recommended formal guidance confirming that Australian copyright law protects works created by humans using AI tools where there is sufficient human authorship.
Limit copyright transparency requirements to useful rights-management/accountability information without commercially or technically sensitive detail.
Transparency requirements should therefore focus on information that is genuinely useful for rights management and accountability without requiring disclosure of commercially or technically sensitive detail.
Support continued specialist copyright work through the Attorney-General's portfolio rather than a separate and potentially inconsistent copyright regime within broader AI legislation.
AIIA recommends that the Committee support continued specialist work through the Attorney-General's portfolio rather than create a separate and potentially inconsistent copyright regime within broader AI legislation.
Combine fraud/identity-misuse enforcement with platform safeguards, provenance/authenticity technology, rapid reporting/removal and public education.
Responses should combine enforcement against fraud and identity misuse with platform and service safeguards, provenance and authenticity technologies, rapid reporting and removal pathways, and public education about evolving attack methods.
Encourage interoperable authenticity standards while preserving satire, accessibility, creative production and privacy-preserving editing.
Policy should encourage interoperable authenticity standards while preserving legitimate uses such as satire, accessibility, creative production and privacy-preserving editing.
Coordinate AI safety obligations with online safety, Digital Duty of Care, privacy, consumer and sector regimes to avoid duplication or conflict.
AIIA supports coordinated design so that any AI-related safety obligations complement, rather than duplicate or conflict with, the Online Safety Act, the Digital Duty of Care, privacy, consumer protection and sector-specific requirements.
Maintain existing critical-infrastructure and sector cyber regimes where applicable; clarify their extension to model services, agents and AI supply chains rather than automatically adding a separate regime.
The Security of Critical Infrastructure framework and sector-specific cyber requirements should remain the primary regulatory architecture where they already apply. AI-specific guidance should clarify how existing obligations extend to model services, agentic systems and AI supply chains rather than automatically create a separate compliance regime.
When immediate legacy replacement risks safety, continuity or operations, permit segmentation/compensating controls under approved transition plans; focus enforcement on unmanaged material risk, with guidance and realistic timeframes.
Where immediate replacement would create safety, continuity or operational risks, entities should be able to use segmentation and other compensating controls under an approved transition plan. Enforcement should focus on unmanaged material risk rather than the age of technology alone, supported by clear guidance and realistic implementation timeframes.
Coordinate intelligence, law enforcement, cyber, electoral and online-safety responses to AI interference; focus on attribution, detection, platform integrity, resilience and consequences while protecting legitimate political communication and research.
These risks require coordination across intelligence, law enforcement, cyber-security, electoral and online-safety bodies. Responses should focus on attribution, detection, platform integrity, public resilience and consequences for malicious activity while protecting legitimate political communication and research.
Keep the AI Safety Institute from duplicating general regulators; ordinarily retain statutory agencies’ decisions and enforcement, with Institute work covering measurement, emerging-risk analysis, technical coordination and international interoperability.
AIIA supports the Institute as a technical evaluation and evidence capability rather than a duplicative general regulator. Its comparative advantage should be model evaluation, measurement, emerging-risk analysis, coordination of technical expertise and international interoperability. Regulatory decisions and enforcement should ordinarily remain with the agencies that hold the relevant statutory mandate.
Have the Institute engage industry and researchers; make evaluation methods transparent and internationally aligned where possible to avoid unnecessary duplicate testing.
The Institute should also maintain structured engagement with industry and researchers so that government receives timely information about emerging capabilities, mitigations and deployment practices. Where possible, evaluation methods should be transparent and aligned with international work so Australian organisations are not required to undergo unnecessary duplicate testing.
Make compute-assurance requirements proportionate to workload sensitivity and consequence, including stronger physical, personnel, redundancy, supply-chain and cyber controls for sensitive workloads.
The policy framework should recognise that different workloads carry different risks. General commercial cloud workloads may require one level of assurance, while sensitive workloads may require stronger physical security, personnel assurance, redundancy, supply-chain integrity and cyber controls. Requirements should be proportionate to the sensitivity and consequence of the workload.
Integrate data-centre growth into energy, grid, transmission, water and land-use planning, addressing local impacts transparently.
Data-centre growth must be integrated into energy, grid, transmission, water and land-use planning. AIIA recognises community concerns about electricity prices, water use, location and local benefits. These issues should be addressed directly and transparently through coordinated planning.
Use outcomes-based, nationally consistent data-centre settings that avoid duplicating state planning, accommodate technological change and offer transparent approval/compliance pathways.
AIIA supports the objective of national consistency and clear investment signals. The detailed framework should be outcomes-based, avoid duplicating state planning processes, accommodate technological change and provide transparent pathways to approval and compliance.
Where major data-centre projects receive public funding, concessional finance or material taxpayer-funded infrastructure, negotiate proportionate Australian research/innovation access commitments reflecting technical capabilities, security requirements and operating model.
Where a major data-centre project receives public funding, concessional finance or material taxpayer-funded enabling infrastructure, government should negotiate proportionate commitments to Australian research and innovation access. These commitments should reflect the facility’s technical capabilities, security requirements and operating model.
Avoid uniform compute quotas on every planning approval; guarantee suitable compute access without artificially separating domestic and global capability.
A uniform quota attached to every planning approval would be ineffective. Compute capacity varies considerably, and a fixed reservation could deter investment or create capacity that eligible Australian users cannot practically use. Australia should be guaranteeing access to suitable compute. Australia should avoid policies that artificially separate domestic and global capability.
Set clear sustainability outcomes and allow flexible industry compliance rather than potentially obsolete technology-specific requirements.
Government should set clear outcomes and allow industry flexibility to meet them rather than locking in technology-specific requirements that may become obsolete.
Keep open-standards requirements proportionate, protecting cyber security, privacy, commercial confidentiality and IP; do not require proprietary source code, model weights or security-sensitive information.
These requirements should be proportionate and should protect cyber security, privacy, commercial confidentiality and intellectual property. Open standards should not be interpreted as requiring suppliers to disclose proprietary source code, model weights or security-sensitive information.