Reference
What regulators have actually said about AI
Between late 2024 and the middle of 2026, regulators and professional bodies across Australia, the United Kingdom and the United States published guidance on using AI in professional work. None of them banned it. Read together, they say four things — in strikingly similar terms.
- Documents read at the issuer
- 13
- Quotations, each dated and linked
- 22
- Jurisdictions
- 3
- Widely repeated quotations dropped, because the regulator never said them
- 4
How to read this page
Every quote here was read at the source
13 documents, each opened at the issuing body itself rather than in somebody's summary of it. Where a widely repeated quotation turned out to be a paraphrase written by a vendor or a journalist, it was removed rather than tidied up — that happened four times while this page was being built.
Dates are part of the quote. One of these instruments was amended after it was first issued and is softer now than the version still circulating; one obligation below has not started yet; and one record-keeping rule quoted here predates the AI guidance entirely and is not about AI at all. Where that matters, it is written on the card.
Point 1 of 4
The duties you already had attach to the AI output
None of these created a new rulebook. Each says the obligations already in force apply to work done with AI, in the same terms.
tax practitioners are still ultimately responsible for the tax agent services they provide to their clients. It is important for tax practitioners to understand the capabilities and limitations of AI tools, and ensure AI outputs are assessed and supplemented by professional judgement, before being relied on.
The regulatory framework for financial services and credit is technology neutral. Licensees need to consider their existing regulatory obligations before deploying AI.
Regardless of what technology is used in providing healthcare the practitioner remains responsible for delivering safe and quality care and for ensuring their own practice meets the professional obligations set out in their Code of Conduct.
ethical obligations of competence, diligence, and confidentiality remain unchanged.
Point 2 of 4
Someone has to check the output
Verification appears in every one of them, and in two it is said explicitly that the checking cannot be delegated back to the same kind of tool.
Tax practitioners should verify and review AI generated content for accuracy throughout each step of the workflow and should establish processes to understand and contest AI decisions/outputs. Each of these steps should be documented.
solicitors should not rely on generative AI to verify sources produced by AI.
If using an AI scribing tool, the practitioner is responsible for checking the accuracy and relevance of records created using generative AI.
Such verification must not be solely carried out by using a Gen AI tool or program.
Point 3 of 4
Responsibility does not move
The practitioner, licensee, or firm stays accountable for the result, whatever produced it. Nothing here treats a tool as a party that can carry blame.
Taxpayers remain accountable for ensuring the information they or their agents provide to the ATO is accurate – whether the advice came from a friend, online sources, or if AI tools were used in its preparation.
Licensees must consider their existing obligations and duties when it comes to the deployment of AI and avoid simply waiting for AI laws and regulations to be introduced.
Do not upload client data into a public-facing AI tool like ChatGPT. Uploading client data into in-house AI tools is acceptable, as long as they possess adequate privacy and security attributes.
Point 4 of 4
Be able to show what happened
The artefact being asked for is shifting from a policy that says what a firm would do, to a record of what it actually did.
You must keep records that correctly record the tax agent services you have provided, or that are provided on your behalf, to each of your clients, including former clients.
As a mortgage broker, we expect you to keep records of how you have acted when providing credit assistance. This includes records of inquiries you make into the consumer's circumstances, and the consideration, investigation and assessment of the products you recommend.
consider maintaining records of each generative AI output for discovery and retention purposes.
Document AI usage and verification processes.
The most repeated position
Client information in a public AI tool
Four bodies in two countries reached the same view, in their own words, without coordinating. The line they draw is between a public tool and a controlled one — not between using AI and not using it.
any placing of client confidential information into a publicly available generative AI system is akin to putting it in the public domain. This is likely to be a breach of confidentiality and clients may lose privilege.
Lawyers cannot safely enter confidential, sensitive or privileged client information into public AI chatbots/copilots (like ChatGPT), or any other public tools.
The input of client data into publicly available AI tools is likely to constitute a breach of client confidentiality, unless the client has consented to this.
Do not upload client data into a public-facing AI tool like ChatGPT. Uploading client data into in-house AI tools is acceptable, as long as they possess adequate privacy and security attributes.
Dated, and not yet started
One obligation with a date on it
Worth knowing about in advance, because it is the only thing here with a commencement date rather than a publication date.
From 10 December 2026, APP entities that use personal information in ADM with the potential to affect rights or interests will be required to provide information in their privacy policies about the kinds of personal information used and the kinds of decisions made using ADM.
Sources
Every document quoted on this page
- TPB(GS) 55/2026 — The use of Artificial Intelligence and the Code of Professional ConductTax Practitioners Board · 22 July 2026
- How far can you go with AI before you hit an ethical dilemma? (CPA Australia InPractice)Brendan O'Connell, quoted by CPA Australia · 25 August 2025
- From 'hacks' to half-truths: ATO warns of tax time misinformation and reveals focus areasAnita Challen, Assistant Commissioner, Australian Taxation Office · 27 April 2026
- A Solicitor's Guide to Responsible Use of Artificial IntelligenceThe Law Society of New South Wales · January 2026
- Statement on the use of artificial intelligence in Australian legal practiceThe Law Society of NSW, the Legal Practice Board of Western Australia, and the Victorian Legal Services Board and Commissioner · 6 December 2024
- Practice Note SC Gen 23 — Use of Generative Artificial IntelligenceSupreme Court of New South Wales · issued 28 January 2025, commenced 3 February 2025
- Regulatory Guide 273 — Mortgage brokers: Best interests dutyAustralian Securities and Investments Commission · June 2020, in force from 1 January 2021
- Report 798 — Beware the gap: Governance arrangements in the face of AI innovationAustralian Securities and Investments Commission · October 2024
- Media release 24-238MRJoe Longo, Chair, Australian Securities and Investments Commission · 29 October 2024
- Meeting your professional obligations when using Artificial Intelligence in healthcareAhpra and the National Boards · page reviewed 22 August 2024
- Topical guidance covering the application of PCRT to the ethical use of artificial intelligence toolsThe PCRT Bodies, published by ICAEW · 19 January 2026
- OPR Alert 2026-19 — Introductory Guidelines for Responsible AI Use in Federal Tax PracticeOffice of Professional Responsibility, United States Internal Revenue Service · 24 June 2026
- Consultation on guidance for transparency in automated decision-makingOffice of the Australian Information Commissioner · 18 May 2026, for an obligation commencing 10 December 2026
These are quotations from published documents, reproduced with their dates and sources so you can read them yourself. They are not legal, privacy or compliance advice, and this page is not a statement of what your obligations are. No body quoted here has any relationship with FlowClave, and nothing here means FlowClave makes a firm compliant with anything or satisfies any obligation described. Your firm remains responsible for its own obligations. Several of these documents are under periodic review — check the date, and read the source.
One job, defined once
Bring one job your firm repeats
Thirty minutes on how that job actually runs, and a written map you keep either way.