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 58

Attestor Pty Ltd

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.

Argues that compliance capability and data-handling trust constrain accounting AI adoption more than absent law. Recommends deployment-neutral sovereignty standards, aligned tax and privacy obligations, implementation support and compliance skills. Attestor discloses its commercial interest, allows alternative deployment models, and describes its practice conversations as qualitative and subject to selection bias.

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. Argues that accounting practices already face up to four regulatory obligations and that capability to comply, not absence of law, is the binding constraint on AI adoption.

    Australian accounting practices are not entering an unregulated space when they adopt AI. Up to four obligations already apply, and the capability to comply with them, not the absence of law, is the binding constraint on adoption.

  2. Says data handling is consistently the largest reported barrier to deeper AI use, and that practitioner hesitation is about where client data went and who could reach it.

    Data handling is consistently the largest reported barrier to deeper use, and what practitioners tell us confirms it: the hesitation is not about whether AI works, but about whether they can answer a client's or a regulator's question about where the data went and who could reach it.

  3. Says practices below $3 million turnover are generally outside most Privacy Act obligations, while the Tax File Number Rule applies regardless of turnover.

    Practices with annual turnover under $3 million are generally not APP entities and are outside most of the Privacy Act, although the Tax File Number (TFN) Rule binds all TFN recipients regardless of turnover.

  4. Argues that residency is necessary but not sufficient for data sovereignty; the remainder is knowing which jurisdiction's law can reach the data.

    Residency is therefore necessary but not, on its own, the whole answer; the remainder is knowing which jurisdiction's law can reach the data, and on what terms.

  5. Proposes principles for domain-specific models: a base model deployable in-jurisdiction and a domain layer trained only on public law and regulator material, never client data.

    (a) The base model must be deployable in-jurisdiction; a model that can only be reached through a foreign API is not a sovereign capability. (b) The domain layer is trained on public law and regulator material obtained through published channels only; client data never enters training.

  6. Argues that compliance with several regulatory regimes is a skills problem for the accounting profession before it is a software problem.

    The compliance work described above is a skills problem before it is a software problem: someone in a ten-person practice has to understand four regulatory regimes well enough to map the firm's AI use against them, and that person is usually a partner doing it after hours.

Recommendations

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

  1. Recognise data sovereignty, meaning Australian residency plus freedom from foreign extraterritorial access, as an enabler of SME adoption in regulated professions, with adoption support neutral as to deployment model but specific as to that outcome.

    Recognise data sovereignty, meaning Australian residency together with freedom from foreign extraterritorial access, as an enabler of small and medium enterprise (SME) adoption in regulated professions, and frame adoption support so that it is neutral as to deployment model but specific as to that outcome.

  2. Have the ATO, TPB and Attorney-General’s Department align tax-practitioner AI obligations, resolving small-business-exemption disparities and clarifying third-party status under Code item 6 for common configurations, rather than leaving practitioners to interpret “configured and used” alone.

    Recommend that the Australian Taxation Office (ATO), the TPB and the Attorney- General's Department align the obligations that apply to tax practitioners' use of AI, including: (a) resolving the two-tier effect of the Privacy Act small business exemption across a profession where every practitioner holds equally sensitive data; and (b) clarifying, for the configurations in common use, when an AI tool constitutes a third party for the purposes of item 6 of the Code, so that practitioners are not left to interpret "configured and used" alone.

  3. Have the OAIC accompany its automated decision-making guidance with implementation templates for small entities, and make enforcement after 10 December 2026 education-first for small businesses.

    Recommend that the Office of the Australian Information Commissioner (OAIC) accompany its automated decision-making guidance with implementation templates for small entities, and that enforcement in the period after 10 December 2026 be education-first for small businesses.

  4. Direct small business AI adoption funding toward demonstrated capability to evidence AI-assisted work, rather than toward tool licences alone.

    Direct small business AI adoption funding toward the demonstrated capability to evidence AI-assisted work, rather than toward tool licences alone.

  5. Treat compliance capability as a workforce skills need for the accounting profession, within the Committee's consideration of skills under term of reference (g).

    Treat compliance capability as a workforce skills need for the accounting profession, and include it within the Committee's consideration of skills under term of reference (g).

  6. For small regulated businesses, treat data sovereignty as the ability to state where client data sits and whose law can reach it, with a deployment-neutral standard.

    The Committee should therefore treat "data sovereignty" in term (f), for small regulated businesses, as the ability to state plainly where client data sits and whose law can reach it. That standard is deployment-neutral.

  7. Treat narrow, domain-specific models as a best-practice question as much as a capability question.

    Term (f) also asks about narrow, domain-specific models in areas of Australian sectoral expertise. Australian tax and audit work is such an area. General-purpose models are trained predominantly on non-Australian material and carry no reliable knowledge of Australian tax law, TPB requirements or ATO practice. The Committee should treat this as a best-practice question as much as a capability one.

Read the original

Open in the interactive guide