Submission 351
Name withheld
Submitter group: Individuals.
Position labels are not shown for individual submitters.
An AI model identified the submitter group and chose the extracts. Counts do not measure importance or merit.
Extracts from the submission, in the submitter’s own words.
Extracts
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.
However, embracing AI does not require Australians to surrender privacy, autonomy, intellectual property, procedural fairness or meaningful human oversight.
Government should establish clear requirements concerning the storage, processing and jurisdictional control of sensitive Australian data.
Where AI materially contributes to a government decision affecting an individual, the person should ordinarily have a right to know.
A statement that a "human remains in the loop" is meaningless unless that human has genuine authority and sufficient information to change the outcome.
AI systems should not exploit children's vulnerability through manipulative personalised content, inappropriate profiling or unnecessary collection of behavioural information.
Australian copyright law should provide clarity about when protected work can be used for commercial AI training. The fact that material can technically be accessed online should not automatically mean it is available for unrestricted commercial exploitation.
Extracts from recommendations
Extracts chosen by an AI model, in the submitter’s own words. Listed in the same order as on its page in the interactive guide.
Government should clearly distinguish between information that is publicly available, information supplied for a particular administrative purpose, information that has been commercially collected, and information for which a person has genuinely consented to AI processing. Providing information to a government department should not automatically constitute consent for that information to be repurposed for unrelated AI development or training.
Likewise, accepting the terms of an essential digital service should not become a legal fiction allowing unlimited secondary use of personal information.
Australians should be told, in understandable language: when their personal information is processed using AI; why AI is being used; whether information is used to train or improve a system; whether information leaves Australia; which organisations can access it; how long it is retained; whether it is combined with information from other sources; and what rights a person has to object, correct information or seek review.
Consent should be meaningful rather than buried within lengthy terms and conditions.
Government should establish clear requirements concerning the storage, processing and jurisdictional control of sensitive Australian data. Particular caution should apply to: health information; disability information; children's information; biometric information; government records; legal information; financial information; critical-infrastructure data; and information affecting national security.
Australia should carefully consider the consequences of becoming dependent upon overseas AI providers for essential government, commercial and infrastructure functions. Sensitive Australian information should not automatically be transferred overseas merely because a foreign provider offers the cheapest or most convenient service.
Australia should also invest in sovereign AI capability rather than becoming permanently dependent upon a small number of foreign corporations.
Where AI materially contributes to a government decision affecting an individual, the person should ordinarily have a right to know.
Human review must not consist of an officer merely accepting an algorithmic recommendation. For significant government decisions, the responsible human decision-maker should understand: what the AI system considered; what information it relied upon; the limitations of the system; whether relevant information may be incorrect; why a particular recommendation was generated; and their authority to reject it.
People should have a practical right to request human reconsideration of significant automated or AI-assisted decisions.
Government should never be permitted to defend a harmful decision by saying effectively that the system produced the result. Someone must remain legally accountable. Every significant Commonwealth AI system should have an identifiable department, accountable officer and documented decision-making framework.
I recommend a public register of significant AI systems used by Commonwealth departments and agencies. Subject to legitimate national-security exceptions, it should identify: the agency using the system; its purpose; the provider; whether the system was developed in Australia or overseas; the general categories of information it uses; whether personal information is involved; whether it makes or recommends decisions; the safeguards applying to it; independent assessments undertaken; significant identified failures; and avenues available to affected individuals.
High-impact systems should undergo independent testing before deployment and regular auditing afterwards. Testing should examine: accuracy; bias; false positives and false negatives; cybersecurity; privacy; reliability; explainability; performance across different populations; susceptibility to manipulation; and consequences when the system fails.
Government procurement contracts should provide sufficient access for independent auditing. Commercial confidentiality should not become an excuse preventing scrutiny of technology exercising public power.
Responsibility should not disappear between the developer, provider, deployer and user. The law should clearly establish who is responsible for particular types of AI-related harm. There should also be accessible complaint and review mechanisms that do not require an ordinary person to understand the technical architecture of the AI system that affected them.
AI systems should not exploit children's vulnerability through manipulative personalised content, inappropriate profiling or unnecessary collection of behavioural information.
Particular caution is warranted around: facial recognition; biometric information; educational profiling; behavioural prediction; targeted advertising; synthetic sexual or abusive imagery; deepfakes; and AI systems designed to create emotional dependence.
Law and enforcement should be capable of responding quickly to AI-generated material used for: fraud; scams; identity theft; sexual exploitation; extortion; harassment; reputational harm; and impersonation of public authorities.
Victims also need rapid mechanisms for removal of harmful synthetic material.
Banks, telecommunications providers, digital platforms, AI companies and government all have roles in preventing AI-enabled fraud. The Committee should consider responsibilities concerning detection of synthetic impersonation, fraudulent advertising, mass scam activity and misuse of AI services.
Consumers should not carry the entire financial burden of sophisticated fraud where organisations could reasonably have detected or prevented it.
Australian copyright law should provide clarity about when protected work can be used for commercial AI training. The fact that material can technically be accessed online should not automatically mean it is available for unrestricted commercial exploitation.
Creators deserve transparency about the use of their work.
Parliament should examine workable mechanisms for consent, licensing, attribution and remuneration while preserving legitimate research, innovation and other appropriate uses.
Australia should develop standards for identifying significant AI-generated or materially altered content, particularly in: political communication; government communication; news-like content; commercial advertising; impersonation; financial promotion; and realistic synthetic audio or video. This must be designed carefully so legitimate satire, creativity and ordinary editing are not unnecessarily regulated.
Political parties, campaign organisations and candidates should be subject to clear rules concerning materially deceptive synthetic media.
12. protect Australian elections and public debate from malicious AI impersonation and foreign interference;
Where large AI or data-centre developments materially affect local electricity, water, land or infrastructure, communities should receive transparent information about those impacts and genuine opportunities to participate in planning decisions.
Regional communities should benefit through better healthcare access, agricultural technology, education, business opportunities, government services and employment. They should not simply become locations for energy-intensive infrastructure while the economic benefits flow elsewhere.
Government should require appropriate disclosure of expected electricity and water use for major developments and assess cumulative regional impacts.
Where public infrastructure is upgraded principally to support major private AI infrastructure, Australians should be able to see who pays and who benefits.
Households and small businesses should not quietly subsidise infrastructure costs created by extremely large commercial users.
Australians who cannot or do not wish to interact with an AI system should retain reasonable access to essential government services. Digital transformation should improve access, not eliminate alternatives. Particular consideration is needed for older Australians, people with disability, people experiencing digital exclusion, people with limited connectivity and people who simply need to speak with a human because their circumstances do not fit an automated process. Efficiency should not eliminate accessibility.
Government should monitor where AI causes significant workforce disruption and support retraining and transition.
Australia's education system should also teach people how to work effectively with AI while retaining independent reasoning, verification and professional judgement.
Government should provide practical guidance enabling smaller businesses to adopt useful AI safely without imposing unnecessarily complex compliance obligations.
At the same time, small businesses should receive meaningful protection against unfair contractual terms or data practices imposed by dominant AI providers.
Australia should retain domestic capability in areas important to national resilience, including research, cybersecurity, health, agriculture, defence and government.
Australian researchers and businesses should have pathways to commercialise innovation here rather than being forced to move overseas simply to obtain capital or computing capacity.
Any expansion of intelligence or law-enforcement powers in response to AI should remain necessary, proportionate, legally defined and independently overseen.
Regulation should therefore be proportionate to potential harm. Higher-risk applications should attract stronger requirements concerning testing, transparency, recordkeeping, human oversight and accountability. Low-risk uses should not be buried under unnecessary bureaucracy.