Submission 122
Professor Jeannie Paterson
Submitter group: Individuals.
Position labels are not shown for individual submitters.
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 in a personal capacity that existing consumer law covers much AI harm but enforcement remains limited. Proposes coordinated, distributed regulation, transparency and accountability initiatives, and consideration of pre-market safety obligations and high-risk agentic AI standards. Says transparency and AI literacy need complementary governance and substantive responsibilities, rather than reliance on disclosure alone.
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.
Says existing Australian law applies to a substantial portion of AI concerns, but little enforcement means regulator capacity remains untested.
Existing Australian law applies to a substantial portion of AI concerns. However, as yet, there has been little by way of enforcement and therefore regulator capacity remains untested.
Argues that a supplier does not avoid consumer law liability by attributing the outcome to the training data or the algorithm.
A supplier does not avoid liability by attributing the outcome to the training data or the algorithm
Says consumer protection law contains no requirement for systematic oversight of AI consumer products or for processes supporting explainable and accountable AI.
Consumer protection law contains no requirement for systematic oversight of AI consumer products, and no requirement for processes supporting explainable and accountable AI.
Argues that regulation giving developers and providers incentives to address systemic risk, with penalties, is likely to be more durable than prescriptive rules, which date quickly.
Prescriptive rules can provide an illusion of certainty but have a tendency to date quickly. Regulation that provides incentives for developers and providers to address systemic risk, with penalties for failing to do this, is likely to be more durable.
Argues that AI literacy on its own is insufficient for safe AI use.
AI literacy on its own is insufficient for the goal of safe AI use. I have drawn a comparison with consumer credit regulation.
Says a single AI product may fall within the remit of several regulators, with none taking overall responsibility.
Falling between the regimes: a single AI product may fall within the remit of several regulators, with none taking overall responsibility.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Consider ex ante (pre-market) safety obligations, given rising risk.
Given rising risk, there may be a case for ex ante (pre-market) safety obligations.
Coordinate regulators and understand the inter-relationship between laws as a regulatory network.
Effective legal regulation to promote safe and responsible AI requires coordination between regulators and an understanding of the inter-relationship between laws or, in other words, the regulatory network.
Treat transparency and AI literacy as areas for government initiatives.
Transparency and AI literacy are necessary preconditions to effective regulation, and ripe for government initiatives
Extend AI literacy programs to ethics and responsibilities.
Literacy programs that extend to AI ethics and responsibilities are clearly essential to benefit from the workforce efficiencies of AI.
Pursue transparency and accountability initiatives for AI-enabled consumer services, through existing or new law.
These insights support initiatives for transparency and accountability in AI-enabled consumer services, either through existing or new law.
Consider international safety red lines and outcome-style monitoring of locally trained and deployed foundation models as concerns about high-risk agentic AI and AI agents rise.
As concerns about high-risk agentic AI and AI agents rise, there is a stronger case for international standards on safety ‘red lines’ and outcome-style safety monitoring of locally trained and deployed foundation models.
Keep outsourced assurance risks in view when considering systems certified under the EU AI Act.
Outsourced assurance: in Gill v Ethicon, the Therapeutic Goods Administration conducted no independent assessment of safety or efficacy where products carried a CE mark and required no evidence of independent pre-market testing.30 This issue should be kept front of mind for systems certified under the EU-AI Act.