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 250

FinTech Australia

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.

Open in the interactive guide

Summary

Read the original submission for the full text.

FinTech Australia, the peak body for the fintech sector, argues for a proportionate, technology-neutral approach to AI regulation, with new rules directed at demonstrated gaps in existing law. It addresses SME adoption, AI-generated financial advice, tax agent services rules, agentic commerce and payments, and the Scams Prevention Framework.

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. Says the productivity gains from AI will depend heavily on adoption by small and medium-sized businesses.

    The productivity gains from AI will depend heavily on adoption by small and medium-sized enterprises.

  2. Recommends checking whether existing law covers AI harms, can be applied in practice, and gives regulators adequate powers and resources before creating new AI-specific obligations.

    Before creating new AI-specific obligations, the Government should establish whether the relevant harm is already covered by law, whether the existing law can be applied in practice and whether a regulator has the powers and resources needed to enforce it.

  3. Says the effect of the Government's existing privacy reforms should be considered before introducing additional AI-specific transparency requirements.

    The consultation does, however, reinforce the case for considering the effect of the Government’s existing privacy reforms before introducing additional AI-specific transparency requirements.

  4. Argues that equivalent services and risks should attract equivalent obligations, regardless of technology, provider location or description.

    The law should apply according to the service being provided and the risk it presents. The technology used to generate the service, the provider’s location and the way the service is described should not permit equivalent conduct to escape equivalent obligations.

  5. Says detailed regulation of agentic commerce would be premature while business models and technical arrangements remain unsettled.

    Our submission considers the emergence of agentic commerce. AI agents may search for products, compare offers and initiate payments on behalf of consumers and businesses. Further work is needed on the application of existing rules to authority, consent, transaction records, liability and dispute resolution. Detailed regulation would be premature while business models and technical arrangements remain unsettled.

  6. Says it is concerned that automatic compensation for scam losses below $3,000 could be exploited, and recommends proportionate verification and safeguards against duplicate or fraudulent claims.

    FinTech Australia is also concerned that the proposed automatic compensation process for losses below $3,000 could itself be exploited by scammers. Any streamlined process should retain proportionate verification and safeguards against duplicate or fraudulent claims, including claims prepared or supported using AI.

Recommendations

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

  1. Treat responsible AI adoption and diffusion as part of the productivity agenda, with attention to regulatory uncertainty and legacy requirements that prevent SMEs accessing useful AI-enabled products and services.

    The Australian Government should treat responsible AI adoption and diffusion as part of its productivity agenda, with particular attention to regulatory uncertainty and legacy requirements that prevent SMEs from accessing useful AI-enabled products and services.

  2. Maintain a proportionate, technology-neutral, risk-based approach to AI regulation, with economy-wide intervention only at demonstrated gaps that existing laws or targeted reform cannot adequately address.

    The Australian Government should maintain a proportionate, technology-neutral and risk-based approach to AI regulation, with economy-wide intervention directed only at demonstrated regulatory gaps that cannot adequately be addressed through existing laws or targeted reform.

  3. Undertake a coordinated, economy-wide review to identify genuine gaps in how existing laws apply to AI, areas of uncertainty or overlap, and legacy provisions that unnecessarily inhibit beneficial AI adoption.

    The Australian Government should undertake a coordinated, economy-wide review to identify genuine gaps in the application of existing laws to AI, areas of uncertainty or regulatory overlap, and legacy provisions that unnecessarily inhibit beneficial AI adoption.

  4. Allow the current privacy reform process and the OAIC's work on automated decision-making to proceed, and assess their operation, before considering additional AI-specific transparency requirements.

    The Australian Government should allow the current privacy reform process and the OAIC’s work on automated decision-making to proceed, and assess the operation of those measures, before considering additional AI-specific transparency requirements.

  5. Any AI-specific transparency requirements should give clear and proportionate information about automated decisions and the use of personal information, while protecting commercially sensitive information and the operational integrity of fraud, security and risk-management systems.

    The Australian Government should allow the current privacy reform process and the OAIC’s work on automated decision-making to proceed, and assess the operation of those measures, before considering additional AI-specific transparency requirements. Any requirements should provide clear and proportionate information about automated decisions and the use of personal information, while protecting commercially sensitive information and the operational integrity of fraud, security and risk-management systems.

  6. Apply the principle that comparable activities and risks attract comparable regulatory obligations, whether delivered by a conventional business, a fintech or a general-purpose AI provider.

    Governments and regulators should apply the principle that comparable activities and risks attract comparable regulatory obligations, regardless of whether a service is delivered by a conventional business, a fintech or a general-purpose AI provider.

  7. Have the Government and ASIC clarify and consistently enforce financial services laws where an AI provider generates financial advice, with a level playing field between offshore AI models and local licensees.

    The Australian Government and ASIC should clarify and consistently enforce the application of the financial services laws where an AI provider generates financial advice for Australian consumers. A level playing field should exist between offshore AI models and locally domiciled financial services businesses.

  8. Have ASIC give licensees practical guidance on AI use across advice, communications, distribution, compliance and service delivery, distinguishing factual information, general advice, decision-support tools and regulated advice.

    ASIC should provide practical guidance to Australian financial services licensees on the use of AI across financial advice, customer communications, product distribution, compliance and service delivery. The guidance should distinguish between factual information, general advice, decision-support tools and regulated financial advice.

  9. Review and amend tax agent services rules to permit bounded automated prompts, explanations and practical assistance through cloud accounting and other digital platforms; distinguish low-risk assistance from advice materially determining or affecting customers' tax positions.

    The Australian Government should review and amend the tax agent services framework to permit appropriately bounded automated prompts, explanations and practical assistance through cloud accounting and other digital platforms. The framework should distinguish low-risk automated assistance from advice that materially determines or affects a customer’s position under a taxation law.

  10. Continue engaging industry on agentic commerce and coordinate with the Reserve Bank's Review of Payments System Regulation, including authority, consent, transaction records, liability and dispute resolution.

    The Australian Government should continue engaging with industry on the development of agentic commerce and coordinate this work with the Reserve Bank’s Review of Payments System Regulation, including its consideration of authority, consent, transaction records, liability and dispute resolution.

  11. Consider AI-specific payments regulation only where a clear gap in the existing framework has been identified that guidance, existing powers or targeted reform cannot address.

    AI-specific payments regulation should only be considered where a clear gap in the existing framework has been identified and cannot be addressed through guidance, existing regulatory powers or targeted reform.

  12. Monitor agentic commerce's effect on competition, payment choice and merchant costs, intervening further only where evidence shows existing frameworks are inadequate.

    The Australian Government should monitor the effect of agentic commerce on competition, consumer payment choice and merchant costs, and only consider further intervention where evidence shows that existing regulatory frameworks are inadequate.

  13. Coordinate any government work on agentic commerce with payments-system reform, Consumer Data Right action initiation, Digital ID and the Financial Innovation Strategy.

    Any government work on agentic commerce should be coordinated with payments-system reform, Consumer Data Right action initiation, Digital ID and the Financial Innovation Strategy.

  14. Ensure Scams Prevention Framework implementation supports responsible use of AI and automated systems to detect and disrupt scams, including timely, reciprocal information sharing between regulated sectors.

    The Australian Government should ensure that the implementation of the Scams Prevention Framework supports the responsible use of AI and automated systems to detect and disrupt scams, including through timely and reciprocal information sharing between regulated sectors.

  15. Give customers useful information and access to review under AI-enabled scam and fraud detection transparency requirements, while protecting the operational integrity of fraud-detection and security systems.

    Transparency requirements applying to AI-enabled scam and fraud detection should provide affected customers with useful information and access to review, while protecting the operational integrity of fraud-detection and security systems.

  16. Resolve low-value complaints efficiently, subject to basic fraud controls.

    Low-value complaints should still be resolved efficiently, subject to basic fraud controls.

  17. Include proportionate verification, coordination between regulated entities and safeguards against duplicate or fraudulent claims in any streamlined Scams Prevention Framework compensation process, including claims generated or supported using AI.

    Any streamlined compensation process under the Scams Prevention Framework should include proportionate verification, coordination between regulated entities and safeguards against duplicate or fraudulent claims, including claims generated or supported using AI.

  18. Use the Office of AI and the Financial Innovation Committee to coordinate AI issues in financial services, identify the agency responsible for each issue and give industry a channel to raise barriers and uncertainty.

    The Australian Government should use the Office of AI and the Financial Innovation Committee to coordinate consideration of AI-related issues in financial services, identify the agency responsible for resolving each issue and provide a clear channel for industry to raise regulatory barriers and areas of uncertainty.

  19. Consider AI when implementing the Financial Innovation Strategy, particularly where existing regulatory uncertainty prevents financial services businesses deploying otherwise useful technology.

    The Government’s Financial Innovation Strategy recognises that financial innovation can strengthen Australia’s economic infrastructure and improve productivity. AI should be considered as this work is implemented, particularly where uncertainty in existing regulation prevents financial services businesses from deploying otherwise useful technology.

  20. Consider how new and existing requirements affect SME access to AI through existing products; address unintended barriers by targeting the rule responsible.

    Government should consider the effect of new and existing requirements on the ability of SMEs to access AI through the products they already use. Where an existing rule creates an unintended barrier, the response should be directed to that rule.

  21. Before adding AI-specific obligations, establish whether existing law covers the harm, can be applied in practice, and gives regulators adequate powers and resources.

    Before creating new AI-specific obligations, the Government should establish whether the relevant harm is already covered by law, whether the existing law can be applied in practice and whether a regulator has the powers and resources needed to enforce it.

  22. Ensure the regulatory review identifies material AI harms or risks inadequately covered by law, unclear or unenforceable laws, regulatory duplication/inconsistency, and legacy automation barriers lacking commensurate benefit.

    The Government should undertake a structured review of the existing framework. The review should identify: ● conduct involving AI that causes or creates a material risk of harm but is not adequately covered by existing law; ● laws whose application to AI is uncertain or difficult to enforce; ● duplication or inconsistency between regulators; and ● legacy requirements that prevent useful forms of automation without delivering a commensurate benefit.

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

  23. Base the regulatory review on actual use cases and distinguish gaps in law from enforcement gaps.

    The review should be practical and based on actual use cases. It should also distinguish between a gap in the law and a gap in enforcement.

  24. Allow decision-category disclosures when they clearly explain personal-information use; avoid disclosure of proprietary methods or thresholds when it would weaken controls or security.

    Businesses should be able to describe categories of decisions, such as credit assessment, identity verification or transaction monitoring, where that provides a clear account of how personal information is being used. There must also be limits on the disclosure of sensitive operational information. Detailed information about fraud detection, scams prevention, cybersecurity, anti-money laundering controls and transaction monitoring may help criminals test and circumvent those systems. Businesses should not be required to disclose proprietary methods, decision thresholds or other information where disclosure would weaken a control or create a security risk.

  25. Allocate transparency obligations according to each party's information and control, recognising third-party AI technology providers.

    The design of the privacy reforms should also recognise the role of third-party technology providers. A fintech may use an AI tool supplied or operated by another entity and may not have access to every technical detail about its operation. Transparency obligations should be allocated according to the information and control held by each party.

  26. Absent legislative change, apply ASIC's finfluencer approach to AI-generated communications: providing financial services as a business without an AFS licence is an offence unless an exemption applies.

    ASIC already publishes guidance for finfluencers. That guidance explains that carrying on a business of providing financial services without an AFS licence is an offence unless an exemption applies.4 In the absence of legislative change, ASIC should take the same approach where the relevant communication is generated through an AI service.

  27. Have ASIC examine general-purpose AI responses to Australians' financial-product prompts, testing whether advice is provided, whether it is personal or general, and whether providers conduct an Australian financial services business.

    ASIC should examine how general-purpose AI services respond to prompts from Australian consumers about financial products. This work should consider whether providers are giving financial product advice, whether the advice is personal or general, and whether the provider is carrying on a financial services business in Australia.

  28. Where existing law applies to AI financial advice, enforce or amend it to create a level playing field for Australian financial services businesses.

    ASIC should examine how general-purpose AI services respond to prompts from Australian consumers about financial products. This work should consider whether providers are giving financial product advice, whether the advice is personal or general, and whether the provider is carrying on a financial services business in Australia. Where the existing law applies, it should either be enforced or amended in such a way which creates a level playing field with Australian financial services businesses.

  29. If ASIC considers existing provisions ineffective, identify precise legal or jurisdictional barriers for possible targeted amendments, avoiding further disadvantage to compliant Australian providers.

    This work should consider whether providers are giving financial product advice, whether the advice is personal or general, and whether the provider is carrying on a financial services business in Australia. Where the existing law applies, it should either be enforced or amended in such a way which creates a level playing field with Australian financial services businesses. If ASIC considers that the existing provisions cannot be applied effectively, it should identify the precise legal or jurisdictional barrier. Government can then consider a targeted amendment. Care should be taken to avoid putting compliant Australian providers at a further disadvantage.

  30. Include financial guidance on factual summaries, user-selected assumptions, customer-based product suggestions, adviser-reviewed material and controls preventing general information services from producing personal recommendations.

    Guidance should address common situations, including: ● an AI tool summarising factual information about a financial product; ● a tool explaining the effect of different user-selected assumptions; ● an assistant suggesting products based on information supplied by a customer; ● the use of AI to prepare material for review by an authorised adviser; and ● controls for preventing a general information service from producing personal recommendations.

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

  31. Keep guidance technology-neutral, describing conduct and controls rather than endorsing models; develop it with firms of different sizes to make expectations workable for fintechs and larger institutions.

    The guidance should remain technology-neutral. It should describe the relevant conduct and the controls expected of the licensee, rather than endorse particular models or system designs. ASIC should develop it with licensed firms of different sizes so that the resulting expectations are workable for fintechs as well as larger institutions.

  32. Make permitted automated tax assistance transparent while preserving suitable protections for advice materially affecting customers' tax positions.

    Second, legacy regulation can prevent businesses from using AI in low-risk and beneficial ways. Requirements under the tax agent services framework can inhibit cloud accounting platforms from providing customers with automated prompts, explanations and practical tips. The framework should be modernised to permit appropriately bounded and transparent digital assistance, while preserving suitable protections for advice that materially affects a customer’s tax position.

  33. Distinguish automated assistance from substantive tax advice, considering whether a feature flags existing records or explains general requirements and matters for customers to check.

    The framework should distinguish between automated assistance and substantive tax advice. Relevant factors could include whether the feature: ● alerts a customer to information already contained in its records; ● explains a general requirement or identifies a matter for the customer to check;

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

  34. For tax-assistance boundaries, consider customer responsibility, professional-advice referrals for material/uncertain issues, confirmation before lodging/altering/approving information, and testing, monitoring and escalation controls.

    makes clear that the customer remains responsible for the decision; ● recommends professional advice where the issue is material or uncertain; ● does not lodge, alter or approve information without the customer’s confirmation; and ● is subject to testing, monitoring and escalation controls. A prompt that an invoice may have been entered twice presents a different risk from a recommendation about how a business should structure a transaction for tax purposes. The law should recognise that difference.

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

  35. Consider a targeted exclusion or other legislative mechanism for defined low-risk automated tax assistance, based on service and safeguards; exclude general software-provider exemptions and substantive tax advice outside the practitioner regime.

    This could be addressed through a targeted exclusion or other legislative mechanism for defined forms of low-risk automated assistance. The exclusion should be based on the nature of the service and the safeguards applied. It should not provide a general exemption for software providers or allow substantive tax advice to be supplied outside the tax practitioner regime.

  36. Develop automated tax-assistance reform with the TPB, ATO, tax practitioners, small business representatives and digital service providers.

    It should not provide a general exemption for software providers or allow substantive tax advice to be supplied outside the tax practitioner regime. The reform would complement the Government’s productivity objectives. Better use of accounting data can help businesses detect mistakes earlier, understand their position and seek professional assistance when it is needed. It would also provide a practical example of updating a legacy regime after identifying how it impedes a beneficial use of AI. The Government should develop the reform with the TPB, the ATO, tax practitioners, small business representatives and digital service providers.

  37. Use the RBA review and related government processes to examine agentic commerce, involving payment providers, merchants, fintechs, consumer groups and AI developers.

    It is too early to know which models will achieve broad adoption. The role of the agent, the degree of customer involvement and the way payments are initiated may differ considerably between services. These differences will affect how existing payments, financial services, privacy and consumer laws apply. Further work is warranted, but detailed regulation would be premature. The immediate task should be to understand how emerging models operate and identify where existing rules may be unclear or incomplete. This work should occur through the RBA’s current review and related government processes, with input from payment providers, merchants, fintechs, consumer groups and AI developers.

  38. When mapping existing rules before further measures, consider recording and communicating customer authority, adjustable agent limits, suspension or withdrawal, instruction records, and disputes among customers, agents, merchants and payment providers.

    Relevant questions may include: ● how the customer’s authority is recorded and communicated; ● whether the customer can set and change limits on the agent; ● how that authority can be suspended or withdrawn; ● what records are retained of the instructions given to the agent; and ● how disputes involving the customer, agent, merchant and payment provider are handled. Existing legal and contractual arrangements may address some of these matters. Their operation will depend on the functions performed by each participant and the design of the service. Government and regulators should map the existing framework before determining whether further measures are required.

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

  39. Monitor payment-method selection transparency, commercial influences, merchants' control over accepted methods and reasonable participation by competing providers; retain existing competition, consumer and payments regulation as the starting point.

    These issues should be monitored as the market develops. Particular attention may need to be given to whether: ● consumers understand how an agent selects a payment method; ● commercial relationships affect that selection; ● merchants retain reasonable control over the payment methods they accept; and ● competing payment providers can participate on reasonable terms. Existing competition, consumer and payments regulation should remain the starting point.

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

  40. Monitor international standards and commercial trials while retaining flexibility for local market conditions.

    International standards and commercial trials will also be relevant, particularly given the cross-border nature of online commerce. Australia should monitor those developments while retaining flexibility to respond to local market conditions.

  41. Allow relevant scam intelligence to be shared in forms usable by detection systems, subject to privacy and security safeguards.

    The SPF should accommodate the responsible use of AI and other automated detection tools. In particular, the rules governing information sharing should allow relevant scam intelligence to be exchanged in a form that regulated entities can incorporate into their detection systems, subject to privacy and security safeguards.

  42. Preserve incentives for businesses to develop and deploy effective prevention technology, without revisiting detailed Scams Prevention Framework design in this inquiry.

    We are not asking the Committee to revisit the design of the SPF in detail as part of this inquiry. The broader lesson is that regulation should preserve incentives for businesses to develop and deploy effective prevention technology.

  43. Protect personal information, confidential business information and automated-detection indicators from misuse through scam-intelligence access and disclosure arrangements.

    The same issue arises when information is shared between regulated entities or government agencies. Scam intelligence may include personal information, confidential business information and indicators used by automated detection systems. Access and disclosure arrangements should protect that material from misuse.

  44. Keep existing regulators responsible for their mandates; coordinate cross-boundary AI issues, lead-agency allocation and consistent requirements without creating a new AI regulator.

    These issues do not require the creation of a new AI regulator. Existing regulators should remain responsible for applying and enforcing the laws within their mandates. Better coordination is needed to identify issues that cross those boundaries, determine which agency should lead the response and avoid duplicative or inconsistent requirements.

Read the original

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

Open in the interactive guide