Federal Court, Federal Circuit and Family Court, Administrative Review Tribunal, and Commonwealth statutes. A state admission does not import that state’s court AI note into a Commonwealth proceeding.
Commonwealth proceedings follow the federal court or tribunal you are in. The Federal Court has published GPN-AI. The Federal Circuit and Family Court has its own practice direction. The Law Society of New South Wales lists an Administrative Review Tribunal generative-AI practice direction. A New South Wales or Victorian Supreme Court note does not apply merely because that is where the solicitor is admitted.
Relevant to A16. Catalogue links: draft a mediation or family-dispute-resolution position paper; draft the expert brief and review a valuation or family-report draft; turn existing parenting orders into a practical arrangements sheet. This direction does not name those tasks.
Announces publication of GPN-AI on 16 April 2026 following consultation; preceded by the Chief Justice statement of 28 March 2025 and a notice of 29 April 2025.
Caution required for pleadings, submissions, evidence and confidential or protected information.
Explains when disclosure of GenAI use may be required.
Tells self-represented litigants not to use generative AI programs (e.g. ChatGPT, Claude, Copilot, Gemini) to get legal help about whether they have a case they can win, as they are not reliable for that.
Self-represented-litigant facing; distinct from GPN-AI.
29 April 2025, Chief Justice D S Mortimer. · Effective 2025-04-29
Notice to the Profession dated 29 April 2025 from Chief Justice D S Mortimer on artificial intelligence use in the Federal Court of Australia. It is a notice, not a practice note. It refers to the Chief Justice's media statement of 28 March 2025. It says judges were considering guidelines or a practice note, that the Court would consult, and that it would consider what other courts had done. Until then, practitioners and self-represented litigants were to use generative artificial intelligence consistently with their existing obligations, and to disclose that use if a judge or registrar required it. Comments were invited to AI_Consultation@fedcourt.gov.au by 13 June 2025. The later notice of 16 April 2026 and General Practice Note GPN-AI are different instruments. This notice is guidance.
Guidance notice of 29 April 2025, not a practice note.
Until a later instrument, use generative artificial intelligence consistently with existing obligations.
Disclose that use if a judge or registrar requires it.
Consultation comments were invited by 13 June 2025.
Last updated on the page on 8 October 2026. First published in February 2025. · Effective 2026-10-08
The Fair Work Commission's artificial intelligence transparency statement. It describes the Commission's own use. It is not a direction to parties. The page says the Commission will not use generative artificial intelligence to make decisions under the Fair Work Act 2009 or the Fair Work (Registered Organisations) Act 2009, and that those decisions can only be made by an appropriate human office holder. Stated uses fall into workplace productivity and analytics for insights, across corporate and service-delivery work, including summarising, transcription, redaction and an initial triage of enterprise agreements for staff. Non-generative artificial intelligence is a component of the document search tool. The Commission says it does not currently use generative artificial intelligence in a way the public could interact with, or be significantly affected by, without a human intermediary. An Artificial Intelligence Assessment Committee oversees use. The Executive Director, Legal Performance and Engagement is the accountable official. The statement was first published in February 2025 and was last updated on 8 October 2026. The separate guidance note for parties who use generative artificial intelligence in a case is a different instrument.
The Commission's own-use statement, not a direction to parties.
The Commission will not use generative artificial intelligence to make decisions under the Fair Work Act 2009 or the Fair Work (Registered Organisations) Act 2009.
The public does not currently interact with the Commission's generative artificial intelligence without a human intermediary.
Last updated on the page on 8 October 2026. First published in February 2025.
Examiner guidance. An inventor must be a natural person. A corporation or AI cannot be an inventor. The manual records Stephen L Thaler [2021] APO 5, overturned by a single judge in Thaler v Commissioner of Patents [2021] FCA 879, then restored by the Full Court in Commissioner of Patents v Thaler [2022] FCAFC 62. Special leave was refused in Thaler v Commissioner of Patents [2022] HCATrans 199 on 11 November 2022. Naming a non-human inventor fails the formalities.
The manual says the Full Court finding is the current state of the law on AI inventorship.
Relevant to A23. Catalogue links: prepare a citation list of material prior art; build a prior-art hit table; prepare an issues list on AI-generated works and training data; draft an IP licence or assignment. This manual section does not name those tasks.
OAIC APP Guidelines Chapter 1, updated 30 September 2026. From 10 December 2026, an APP entity must add information to its APP privacy policy if it arranges for a computer program to use personal information to make, or to do a thing substantially and directly related to making, a decision that could reasonably be expected to significantly affect an individual's rights or interests. A law firm that is an APP entity and uses AI in that way needs to update its privacy policy before that date.
"An APP entity must include additional information in its APP Privacy Policy if it arranges for a computer program to use personal information to make, or do a thing that is substantially and directly related to making, a decision, that could reasonably be expected to significantly affect the rights or interests of an individual."
Applies from 10 December 2026 (Privacy and Other Legislation Amendment Act 2024)
Privacy policy must list the kinds of personal information used, decisions made solely by computer programs, and decisions substantially assisted by them (APP 1.8)
OAIC published an APP 1.7-1.9 fact sheet, a government supplementary fact sheet and a flowchart
listed as advice under development; expected completion to be advised
Applies to the ATO's advice-under-development list, not to a ruling that has been issued. Administration-issues item 4296, Taxpayer use of artificial intelligence in tax affairs, says the ATO is proposing to update MT 2008/1 on reasonable care and MT 2008/2 on a reasonably arguable position where taxpayers use AI, and to prepare a fact sheet called ATO statement on the use of artificial intelligence in tax affairs. Expected completion is to be advised. The administration-issues heading says last updated 5 October 2026. That date is the list heading.
The proposed fact sheet is described as practical information about using accurate facts, checking outputs against legislation, case law and ATO public advice, and reviewing outputs before reliance.
Relevant to A26. Catalogue links: reconcile a tax computation; prepare an information request for a return position; build a tax due-diligence issues list. This list does not name those tasks.
TPB webinar page, issued 25 August 2026, with Q&A on TPB(GS) 55/2026. AI outputs are a starting point, the practitioner remains responsible, and client permission is needed before client information goes into an AI tool that is a third party. This page sits beside the guidance statement already in the catalogue.
"Tax practitioners must obtain permission from their client, before disclosing any information relating to a client's affairs to a third-party. This may include entering client information into AI models and tools."
A third party under the Tax Agent Services Act 2009 "can include AI models and tools depending on how these tools are configured and used"
"AI outputs should be treated as a starting point only"; practitioners must review, test and verify before relying on them
Keep enough records to show AI outputs were reviewed in line with the practitioner's processes
Where AI use is a material change to service delivery or data handling, consider updating engagement letters and telling clients
This is the ACCC's own internal policy on how the agency uses AI. Published in February 2026 and reviewed at least once a year. People at the ACCC remain responsible for oversight and decisions. The stated uses are analytics and workplace productivity, including summaries, transcription and drafts, except documents subject to administrative review. The ACCC does not currently deploy AI in direct public interaction through the Infocentre or ScamWatch. It is not a rule for businesses and it is not advice on the Australian Consumer Law.
It uses the definition of AI in the Policy for the responsible use of AI in government.
The AI accountable official is the General Manager of Data and Intelligence. The Corporate Governance Board has oversight.
Historical background only. This 24 March 2026 media release opened consultation on Exposure Draft information sheet D62/2026 until 21 April 2026. The draft was later finalised as TPB(GS) 55/2026, which is the statement to use.
Exposure draft focuses on competence, confidentiality, supervision and control, and professional judgment
"tax practitioners remain responsible for the services they provide, whether AI tools are used or not"
Office of the Migration Agents Registration Authority
2026-03-05
Applies to registered migration agents. The occupational competency standards page has a section, RMAs and the use of AI tools. It says RMAs may use AI while giving immigration assistance, and remain responsible for inadequate or inaccurate information. The Code of Conduct applies even though AI is not specifically mentioned. Sharing a client's personal details through an AI tool may breach the duty of confidentiality in section 35 of the Code. This page does not state the section 280 offence or a written-consent rule. Last updated 5 March 2026.
Relevant to A30. Catalogue links: draft a client declaration or interview-question list; build a visa-pathway longlist; prepare a response-to-information-request checklist. This page does not name those tasks.
Exploratory scenario pieces by IP Australia's ventures team, published 10 July 2023, on how generative AI may affect patents, trade marks, designs and plant breeder's rights. Not a binding practice rule. The patent manual's AI-inventor section is a separate row.
Scenario pieces on whether traditional understandings of IP hold up with generative AI
Covers design rights, patents, plant breeders' rights and trade marks
Privacy obligations apply to personal information input into a commercially available AI product, and to output that contains personal information. Published 21 October 2024 and updated 17 January 2025. It is for an organisation deploying AI that was built with, collects, stores, uses or discloses personal information, including internal use and free public chatbots. It is not a complete list of every privacy obligation.
Best practice: do not enter personal, especially sensitive, information into public generative AI tools.
A deployer includes internal use and a product or service supplied to others. The page also covers free public chatbots.
The page links three companion PDFs (top five takeaways, a selection checklist and a use checklist). They summarise this guidance.
Relevant to A25, A45, A51. Catalogue links: build a privacy-law gap table; draft a privacy-impact or algorithmic-impact assessment; build an upload checklist; decide whether client material can go into a tool. This guidance does not name those tasks.
For developers of generative AI who are APP entities and who are training or fine-tuning models, including a person who designs, builds, trains, adapts or combines models. Published 21 October 2024 and updated 23 October 2024. Publicly available data is not automatically usable to train or fine-tune a model. Developers must take reasonable steps on accuracy.
Publicly available data is not automatically usable for training.
Sensitive information generally needs consent.
Secondary use of held data for training needs careful APP 6 analysis; seek consent/opt-out if unclear.
APP 10 accuracy duties.
Relevant to A45, A25. Catalogue links: draft a privacy-impact or algorithmic-impact assessment; build a privacy-law gap table. This guidance does not name those tasks.
Review of AI use by 23 AFS and credit licensees, released 29 October 2024. It warns that governance is lagging AI adoption. It reports what ASIC found. It does not add a licence condition.
Media release 24-238MR.
Relevant to advising AFSL/ACL holders.
The review covered 624 use cases as at December 2023, and meetings with 12 of the licensees in June 2024. Only 12 of the 23 licensees had AI policy documents that referenced fairness or related concepts such as discrimination or bias.
The report says the framework is technology neutral. Licensees need to consider existing general licensee obligations, directors' duties and consumer protection provisions, including where a third party develops the model.
Relevant to A28, A3. Catalogue links: build an AI-use gap table against the licence's general obligations; draft a financial-services or credit-licence application narrative. This report does not name those tasks.
Guidance for registered tax and BAS agents on using AI while providing tax agent services, under the Code (section 30-10 of the Tax Agent Services Act 2009) and the 2024 Code Determination. Issued and last modified 22 July 2026. It does not create additional rights or legal obligations beyond the TASA.
Does not create new obligations beyond TASA.
The statement says tax practitioners remain responsible for the services they provide and must assess AI outputs with professional judgement before reliance.
Relevant to A26. Catalogue links: reconcile a tax computation; prepare an information request for a return position; build a tax due-diligence issues list. This statement does not name those tasks.
Guidance on the use of AI in delivering immigration assistanceRegulator guidanceSecondary sourceUnverified dateThis note relies on a secondary report. The primary OMARA page was not retrieved; confirm the primary text before reliance.Secondary. About immigration assistance.
Office of the Migration Agents Registration Authority
Report of the UK-Australia Legal Services Regulatory Dialogue, published by the Legal Services Council, with a section on legal technology and AI comparing regulator, professional-body and court responses in both countries.
Notes rapidly rising use of generative AI by lawyers in the UK and Australia
Summarises the NSW Law Society, LPBWA and VLSB+C joint statement that AI cannot substitute for legal knowledge and must be verified
Notes QLS-led multi-state guidance on AI selection and use
Observes common themes in Australian court practice notes (confidentiality, limits on AI in affidavits, citation verification, expert reports)
Office of the Migration Agents Registration Authority
2026-03-05
OMARA guidance on how AI interacts with lawful immigration assistance, RMA responsibilities and consumer consent.
Only RMAs, Australian legal practitioners and limited exempt persons may give immigration assistance; there is no exemption for AI-generated information
A person who provides immigration assistance through AI may commit an offence under s 280 of the Migration Act
RMAs who use AI remain responsible for the accuracy of the assistance; Code of Conduct standards apply
RMAs need clients' written consent before entering personal details into an AI platform
Consumers can complain to the OMARA about an RMA's use of AI
The page was last updated on 5 March 2026.
Relevant to A30. Catalogue links: draft a client declaration or interview-question list; build a visa-pathway longlist; prepare a response-to-information-request checklist. This guidance does not name those tasks.
OAIC publishes “New AI guidance makes privacy compliance easier for business | OAIC”. New guides for businesses published today by the Office of the Australian Information Commissioner (OAIC) clearly articulate how Australian privacy law applies to artificial intelligence (AI) and set out the regulator’s expectations. Checked 2026-10-10. Read the linked page before relying on it.
New guides for businesses published today by the Office of the Australian Information Commissioner (OAIC) clearly articulate how Australian privacy law applies to artificial intelligence (AI) and set out the regulator’s expectations.
OAIC publishes “Getting AI right benefits businesses, productivity and the community - OAIC”. Strong privacy governance and safeguards are essential for businesses to gain advantage from AI and build trust and confidence in the community. Checked 2026-10-10. Read the linked page before relying on it.
Strong privacy governance and safeguards are essential for businesses to gain advantage from AI and build trust and confidence in the community.
OAIC publishes “Can personal information be used to develop or train GenAI? - OAIC”. We cover when and why privacy should be considered by developers using personal information to train generative AI models, how our new guidance fits in with that being published globally and how we see it continuing to apply as technology changes. Checked 2026-10-10. Read the linked page before relying on it.
We cover when and why privacy should be considered by developers using personal information to train generative AI models, how our new guidance fits in with that being published globally and how we see it continuing to apply as technology changes.
This year, we have seen Australians express increasing concern about the use of their personal information to train generative artificial intelligence (AI) products. The message from the public is clear: organisations need to protect privacy.
Digital AssistantGovernment guidanceGovernmentPrimary sourceConfidence: highProfessional guidance. About A30.
Department of Home Affairs
publication date not stated on the page
Applies to people using the Department of Home Affairs website chatbot. The Digital Assistant uses artificial intelligence-based software to answer visa, citizenship, border and trade questions from the words in the question. It cannot give information on a person's visa or citizenship application. The page tells people to use ImmiAccount or VEVO for status, and to remove personal information such as reference numbers, identity information and contact details. It is not a conduct rule for registered migration agents. The page does not state a publication date.
Relevant to A30. Catalogue link for immigration work such as a visa-pathway longlist. This page does not name those tasks and it does not set a duty for registered migration agents.
The ART President's Gen AI practice direction (dated 20 August 2026), adapted from the Federal Court GPN-AI, setting expectations for everyone in ART proceedings and when Gen AI use must be verified, explained or disclosed.
Adapted from the Federal Court of Australia GPN-AI (16 April 2026) with the Chief Justice's permission
Applies to documents, information, submissions and evidence given to the Tribunal; not to internal Tribunal administration
Sets out when a person must verify, explain or disclose use of Generative AI
Non-compliance may lead to directions to correct or withdraw material, less or no weight, or the Tribunal declining to rely on material
Possible referral of legal or professional representatives to a regulator, usually after the substantive decision
Read with the Common Procedures and Expert Evidence Practice Directions 2026; AI agents may not record case events without permission
FWC web page that puts the GenAI guidance note into user-facing language. From 20 October 2026, a person who uses GenAI to prepare a document must disclose that use, check the document, and say so in the document. Lawyers and other professional representatives must also hyperlink any case law cited. The guidance-note PDF is already in the catalogue. This row is the companion page. The page itself has no revision date.
"These requirements apply to all Commission matters from 20 October 2026."
If GenAI is used to create, write, edit or prepare a document you must tell the Commission and explain how it was used, check it yourself, and state that you completed the checks
Witness statements and declarations must be based on the witness's own knowledge and reflect their own words
"Legal practitioners and other professional representatives also must include hyperlinks to any case law referred to in documents prepared using GenAI."
Non-compliance may mean less weight, costs orders or dismissal; some forms will include a GenAI question
ART Expert Evidence Practice Direction (dated 27 January 2026, commencing 2 March 2026) requiring experts to state whether a report contains Gen AI content and, if so, identify it and certify they checked it.
Experts must state whether the report includes content generated using Generative AI
If it does, they must identify the AI content and the application(s) used
Experts must certify they personally checked all AI content, including cited research, for accuracy and reliability
Cover sheet includes a Gen AI certification checkbox
ART Common Procedures Practice Direction (dated 27 January 2026) that strictly prohibits cameras and audio/video recording or transmission devices, including AI agents, in Tribunal case events without permission.
Para 5.16: cameras and recording/transmission devices "including an Artificial Intelligence (AI) agent" are strictly prohibited in case events without express permission of the Presiding Member or Registrar
ART AI transparency statementTransparency statementTribunalPrimary sourceConfidence: mediumUnverified dateNo publication date in body; table shows target dates of 1 December 2026 for in-development items.Secondary. Administrative Review Tribunal.Administrative & Public Law
Administrative Review Tribunal
Not stated
The ART's statement on its own use of AI under the Commonwealth policy for responsible use of AI in government.
The Tribunal does not, and has no intention to, use AI for its review decision-making function
AI is not used to make operational decisions
Any AI use in service delivery, corporate, policy, legal or fraud domains will be subject to human oversight
Staff must complete mandatory AI training; an internal AI use-case register is in development
The Administrative Appeals Tribunal was replaced by the Administrative Review TribunalNoticeTribunalSecondary sourceConfidence: mediumUnverified dateOn 10 October 2026 a direct request to this page did not return the document. The sentence was returned by a search retrieval of the same URL that day. No separate AAT generative-AI instrument was found. ART instruments already in the catalogue are the successor’s.Secondary. Scope not stated in the stored source.
Administrative Appeals Tribunal
2024-10-14
The Administrative Appeals Tribunal (AAT) was replaced by the Administrative Review Tribunal (the Tribunal) on 14 October 2024.
The Administrative Appeals Tribunal (AAT) was replaced by the Administrative Review Tribunal (the Tribunal) on 14 October 2024. This page contains information relating to the former AAT.
Member Forum Q&A - Artificial IntelligenceStatementExternal dispute schemePrimary sourceConfidence: mediumUnverified dateAFCA describes itself as an ombudsman and external dispute resolution scheme, not a statutory tribunal.Secondary. Scope not stated in the stored source.
Australian Financial Complaints Authority
2026-05-27
AFCA's answers to member questions about rising AI use by complainants and representatives and AFCA's own use of AI.
Generic or AI-generated templates with vague or unsubstantiated allegations are not appropriate
AFCA is monitoring complaint submissions, including from paid representatives
AFCA is exploring AI-enabled tools including transcription, call summaries and efficiency insights
AI does not replace human involvement, especially where decisions are made; complaints are assessed on merits and evidence
CDPP AI Transparency StatementTransparency statementProsecutorCriminalPrimary sourceConfidence: highProfessional guidance. Scope not stated in the stored source.
Office of the Commonwealth Director of Public Prosecutions
2026-02-25
CDPP statement under the DTA Policy for the responsible use of AI in government. It says the CDPP currently does not use AI in public services or advice, or internally, and names the CIO as the accountable official.
"Currently the CDPP does not use AI in any services or advice we provide publicly or currently use AI internally."
Follows the Digital Transformation Agency Policy for the responsible use of AI in government; the CIO is the accountable official, supported by an AI reference group
Any data used in AI systems will be handled under the Privacy Act 1988 and the Protective Security Policy Framework
Statement to be updated as its approach changes, and at least every 12 months
Client-facing fact sheet for people preparing a submission to the Royal Commission into Antisemitism and Social Cohesion, not a conduct rule for lawyers. It says the experiences must be the person's own, output must be checked, and confidential or sensitive information should not go into open AI tools. One answer says data shared with any AI generally cannot be deleted except where terms of service say otherwise. That line is broader than many enterprise terms. Use the page for the client caution only.
AI may help structure material, refine language or summarise key points, but the experiences in a submission must be your own
"You should not put confidential or sensitive information into open AI tools."
Proof-read all AI-generated material; it can include incorrect, false and irrelevant material
Recommends approved private AI tools where sensitive information is involved
Copyright Agency publishes positions and surveys on generative AI and copyright (e.g. market-impact survey July 2026; schools guidance on generative AI tools).
ANAO performance audit, published 29 June 2026, of IP Australia's use of AI in the patent rights process. The ANAO found that use largely effective, and made two recommendations on cyber-security risk and strategic oversight, which IP Australia agreed. This is an audit of the agency, not a conduct rule for IP practitioners.
Published 29 June 2026 as Auditor-General Report No. 43 of 2025-26
IP Australia's use of AI in the patent rights process was largely effective
Two recommendations: cyber-security risk, and strategic oversight of AI implementation. IP Australia agreed
IP Australia deployed its first AI tool into patent examination in 2018
No AI guidance found as of 10 Oct 2026.No guidance foundPrimary sourceOnly submissions found (e.g. "AI: Governance and Leadership"); scrape returned no text. No practitioner guidance.
Law Council of Australia
searched
Only submissions found (e.g. "AI: Governance and Leadership"); scrape returned no text. No practitioner guidance.
No guidance found as of 10 Oct 2026: Australian Bar AssociationNo guidance foundBarPrimary sourceNo standalone ABA AI instrument was found as of 10 Oct 2026. The general news address is a script-rendered list and did not yield a separate ABA practice notice in this review. The Federal Court's 30 April 2025 Notice to the Profession is already recorded against the Court.
Australian Bar Association
Not stated
No standalone ABA AI instrument was found as of 10 Oct 2026. The general news address is a script-rendered list and did not yield a separate ABA practice notice in this review. The Federal Court's 30 April 2025 Notice to the Profession is already recorded against the Court.
No guidance found as of 10 Oct 2026: Copyright Tribunal of AustraliaNo guidance foundTribunalPrimary sourceNo Copyright Tribunal AI practice guidance found as of 10 Oct 2026 (homepage and practice-and-procedure page checked; both minimal). Proceedings run through the Federal Court registry, so the Federal Court GPN-AI may be relevant.
Copyright Tribunal of Australia
Not stated
No Copyright Tribunal AI practice guidance found as of 10 Oct 2026 (homepage and practice-and-procedure page checked; both minimal). Proceedings run through the Federal Court registry, so the Federal Court GPN-AI may be relevant.
This is the catalogue for Commonwealth. An instrument appears under “Rules that apply” only when you name the court or tribunal, or say that no court is involved.
OAIC APP Guidelines Chapter 1, updated 30 September 2026. From 10 December 2026, an APP entity must add information to its APP privacy policy if it arranges for a computer program to use personal information to make, or to do a thing substantially and directly related to making, a decision that could reasonably be expected to significantly affect an individual's rights or interests. A law firm that is an APP entity and uses AI in that way needs to update its privacy policy before that date.
"An APP entity must include additional information in its APP Privacy Policy if it arranges for a computer program to use personal information to make, or do a thing that is substantially and directly related to making, a decision, that could reasonably be expected to significantly affect the rights or interests of an individual."
Applies from 10 December 2026 (Privacy and Other Legislation Amendment Act 2024)
Privacy policy must list the kinds of personal information used, decisions made solely by computer programs, and decisions substantially assisted by them (APP 1.8)
OAIC published an APP 1.7-1.9 fact sheet, a government supplementary fact sheet and a flowchart
listed as advice under development; expected completion to be advised
Applies to the ATO's advice-under-development list, not to a ruling that has been issued. Administration-issues item 4296, Taxpayer use of artificial intelligence in tax affairs, says the ATO is proposing to update MT 2008/1 on reasonable care and MT 2008/2 on a reasonably arguable position where taxpayers use AI, and to prepare a fact sheet called ATO statement on the use of artificial intelligence in tax affairs. Expected completion is to be advised. The administration-issues heading says last updated 5 October 2026. That date is the list heading.
The proposed fact sheet is described as practical information about using accurate facts, checking outputs against legislation, case law and ATO public advice, and reviewing outputs before reliance.
Relevant to A26. Catalogue links: reconcile a tax computation; prepare an information request for a return position; build a tax due-diligence issues list. This list does not name those tasks.
Relevant to A16. Catalogue links: draft a mediation or family-dispute-resolution position paper; draft the expert brief and review a valuation or family-report draft; turn existing parenting orders into a practical arrangements sheet. This direction does not name those tasks.
TPB webinar page, issued 25 August 2026, with Q&A on TPB(GS) 55/2026. AI outputs are a starting point, the practitioner remains responsible, and client permission is needed before client information goes into an AI tool that is a third party. This page sits beside the guidance statement already in the catalogue.
"Tax practitioners must obtain permission from their client, before disclosing any information relating to a client's affairs to a third-party. This may include entering client information into AI models and tools."
A third party under the Tax Agent Services Act 2009 "can include AI models and tools depending on how these tools are configured and used"
"AI outputs should be treated as a starting point only"; practitioners must review, test and verify before relying on them
Keep enough records to show AI outputs were reviewed in line with the practitioner's processes
Where AI use is a material change to service delivery or data handling, consider updating engagement letters and telling clients
This is the ACCC's own internal policy on how the agency uses AI. Published in February 2026 and reviewed at least once a year. People at the ACCC remain responsible for oversight and decisions. The stated uses are analytics and workplace productivity, including summaries, transcription and drafts, except documents subject to administrative review. The ACCC does not currently deploy AI in direct public interaction through the Infocentre or ScamWatch. It is not a rule for businesses and it is not advice on the Australian Consumer Law.
It uses the definition of AI in the Policy for the responsible use of AI in government.
The AI accountable official is the General Manager of Data and Intelligence. The Corporate Governance Board has oversight.
Announces publication of GPN-AI on 16 April 2026 following consultation; preceded by the Chief Justice statement of 28 March 2025 and a notice of 29 April 2025.
Caution required for pleadings, submissions, evidence and confidential or protected information.
Explains when disclosure of GenAI use may be required.
Historical background only. This 24 March 2026 media release opened consultation on Exposure Draft information sheet D62/2026 until 21 April 2026. The draft was later finalised as TPB(GS) 55/2026, which is the statement to use.
Exposure draft focuses on competence, confidentiality, supervision and control, and professional judgment
"tax practitioners remain responsible for the services they provide, whether AI tools are used or not"
Office of the Migration Agents Registration Authority
2026-03-05
Applies to registered migration agents. The occupational competency standards page has a section, RMAs and the use of AI tools. It says RMAs may use AI while giving immigration assistance, and remain responsible for inadequate or inaccurate information. The Code of Conduct applies even though AI is not specifically mentioned. Sharing a client's personal details through an AI tool may breach the duty of confidentiality in section 35 of the Code. This page does not state the section 280 offence or a written-consent rule. Last updated 5 March 2026.
Relevant to A30. Catalogue links: draft a client declaration or interview-question list; build a visa-pathway longlist; prepare a response-to-information-request checklist. This page does not name those tasks.
Tells self-represented litigants not to use generative AI programs (e.g. ChatGPT, Claude, Copilot, Gemini) to get legal help about whether they have a case they can win, as they are not reliable for that.
Self-represented-litigant facing; distinct from GPN-AI.
CDPP AI Transparency StatementTransparency statementProsecutorCriminalPrimary sourceConfidence: highProfessional guidance. Scope not stated in the stored source.
Office of the Commonwealth Director of Public Prosecutions
2026-02-25
CDPP statement under the DTA Policy for the responsible use of AI in government. It says the CDPP currently does not use AI in public services or advice, or internally, and names the CIO as the accountable official.
"Currently the CDPP does not use AI in any services or advice we provide publicly or currently use AI internally."
Follows the Digital Transformation Agency Policy for the responsible use of AI in government; the CIO is the accountable official, supported by an AI reference group
Any data used in AI systems will be handled under the Privacy Act 1988 and the Protective Security Policy Framework
Statement to be updated as its approach changes, and at least every 12 months
29 April 2025, Chief Justice D S Mortimer. · Effective 2025-04-29
Notice to the Profession dated 29 April 2025 from Chief Justice D S Mortimer on artificial intelligence use in the Federal Court of Australia. It is a notice, not a practice note. It refers to the Chief Justice's media statement of 28 March 2025. It says judges were considering guidelines or a practice note, that the Court would consult, and that it would consider what other courts had done. Until then, practitioners and self-represented litigants were to use generative artificial intelligence consistently with their existing obligations, and to disclose that use if a judge or registrar required it. Comments were invited to AI_Consultation@fedcourt.gov.au by 13 June 2025. The later notice of 16 April 2026 and General Practice Note GPN-AI are different instruments. This notice is guidance.
Guidance notice of 29 April 2025, not a practice note.
Until a later instrument, use generative artificial intelligence consistently with existing obligations.
Disclose that use if a judge or registrar requires it.
Consultation comments were invited by 13 June 2025.
Digital AssistantGovernment guidanceGovernmentPrimary sourceConfidence: highProfessional guidance. About A30.
Department of Home Affairs
publication date not stated on the page
Applies to people using the Department of Home Affairs website chatbot. The Digital Assistant uses artificial intelligence-based software to answer visa, citizenship, border and trade questions from the words in the question. It cannot give information on a person's visa or citizenship application. The page tells people to use ImmiAccount or VEVO for status, and to remove personal information such as reference numbers, identity information and contact details. It is not a conduct rule for registered migration agents. The page does not state a publication date.
Relevant to A30. Catalogue link for immigration work such as a visa-pathway longlist. This page does not name those tasks and it does not set a duty for registered migration agents.
Last updated on the page on 8 October 2026. First published in February 2025. · Effective 2026-10-08
The Fair Work Commission's artificial intelligence transparency statement. It describes the Commission's own use. It is not a direction to parties. The page says the Commission will not use generative artificial intelligence to make decisions under the Fair Work Act 2009 or the Fair Work (Registered Organisations) Act 2009, and that those decisions can only be made by an appropriate human office holder. Stated uses fall into workplace productivity and analytics for insights, across corporate and service-delivery work, including summarising, transcription, redaction and an initial triage of enterprise agreements for staff. Non-generative artificial intelligence is a component of the document search tool. The Commission says it does not currently use generative artificial intelligence in a way the public could interact with, or be significantly affected by, without a human intermediary. An Artificial Intelligence Assessment Committee oversees use. The Executive Director, Legal Performance and Engagement is the accountable official. The statement was first published in February 2025 and was last updated on 8 October 2026. The separate guidance note for parties who use generative artificial intelligence in a case is a different instrument.
The Commission's own-use statement, not a direction to parties.
The Commission will not use generative artificial intelligence to make decisions under the Fair Work Act 2009 or the Fair Work (Registered Organisations) Act 2009.
The public does not currently interact with the Commission's generative artificial intelligence without a human intermediary.
Last updated on the page on 8 October 2026. First published in February 2025.
Client-facing fact sheet for people preparing a submission to the Royal Commission into Antisemitism and Social Cohesion, not a conduct rule for lawyers. It says the experiences must be the person's own, output must be checked, and confidential or sensitive information should not go into open AI tools. One answer says data shared with any AI generally cannot be deleted except where terms of service say otherwise. That line is broader than many enterprise terms. Use the page for the client caution only.
AI may help structure material, refine language or summarise key points, but the experiences in a submission must be your own
"You should not put confidential or sensitive information into open AI tools."
Proof-read all AI-generated material; it can include incorrect, false and irrelevant material
Recommends approved private AI tools where sensitive information is involved
Copyright Agency publishes positions and surveys on generative AI and copyright (e.g. market-impact survey July 2026; schools guidance on generative AI tools).
The ART President's Gen AI practice direction (dated 20 August 2026), adapted from the Federal Court GPN-AI, setting expectations for everyone in ART proceedings and when Gen AI use must be verified, explained or disclosed.
Adapted from the Federal Court of Australia GPN-AI (16 April 2026) with the Chief Justice's permission
Applies to documents, information, submissions and evidence given to the Tribunal; not to internal Tribunal administration
Sets out when a person must verify, explain or disclose use of Generative AI
Non-compliance may lead to directions to correct or withdraw material, less or no weight, or the Tribunal declining to rely on material
Possible referral of legal or professional representatives to a regulator, usually after the substantive decision
Read with the Common Procedures and Expert Evidence Practice Directions 2026; AI agents may not record case events without permission
FWC web page that puts the GenAI guidance note into user-facing language. From 20 October 2026, a person who uses GenAI to prepare a document must disclose that use, check the document, and say so in the document. Lawyers and other professional representatives must also hyperlink any case law cited. The guidance-note PDF is already in the catalogue. This row is the companion page. The page itself has no revision date.
"These requirements apply to all Commission matters from 20 October 2026."
If GenAI is used to create, write, edit or prepare a document you must tell the Commission and explain how it was used, check it yourself, and state that you completed the checks
Witness statements and declarations must be based on the witness's own knowledge and reflect their own words
"Legal practitioners and other professional representatives also must include hyperlinks to any case law referred to in documents prepared using GenAI."
Non-compliance may mean less weight, costs orders or dismissal; some forms will include a GenAI question
ART Expert Evidence Practice Direction (dated 27 January 2026, commencing 2 March 2026) requiring experts to state whether a report contains Gen AI content and, if so, identify it and certify they checked it.
Experts must state whether the report includes content generated using Generative AI
If it does, they must identify the AI content and the application(s) used
Experts must certify they personally checked all AI content, including cited research, for accuracy and reliability
Cover sheet includes a Gen AI certification checkbox
ANAO performance audit, published 29 June 2026, of IP Australia's use of AI in the patent rights process. The ANAO found that use largely effective, and made two recommendations on cyber-security risk and strategic oversight, which IP Australia agreed. This is an audit of the agency, not a conduct rule for IP practitioners.
Published 29 June 2026 as Auditor-General Report No. 43 of 2025-26
IP Australia's use of AI in the patent rights process was largely effective
Two recommendations: cyber-security risk, and strategic oversight of AI implementation. IP Australia agreed
IP Australia deployed its first AI tool into patent examination in 2018
ART Common Procedures Practice Direction (dated 27 January 2026) that strictly prohibits cameras and audio/video recording or transmission devices, including AI agents, in Tribunal case events without permission.
Para 5.16: cameras and recording/transmission devices "including an Artificial Intelligence (AI) agent" are strictly prohibited in case events without express permission of the Presiding Member or Registrar
Exploratory scenario pieces by IP Australia's ventures team, published 10 July 2023, on how generative AI may affect patents, trade marks, designs and plant breeder's rights. Not a binding practice rule. The patent manual's AI-inventor section is a separate row.
Scenario pieces on whether traditional understandings of IP hold up with generative AI
Covers design rights, patents, plant breeders' rights and trade marks
ART AI transparency statementTransparency statementTribunalPrimary sourceConfidence: mediumUnverified dateNo publication date in body; table shows target dates of 1 December 2026 for in-development items.Secondary. Administrative Review Tribunal.Administrative & Public Law
Administrative Review Tribunal
Not stated
The ART's statement on its own use of AI under the Commonwealth policy for responsible use of AI in government.
The Tribunal does not, and has no intention to, use AI for its review decision-making function
AI is not used to make operational decisions
Any AI use in service delivery, corporate, policy, legal or fraud domains will be subject to human oversight
Staff must complete mandatory AI training; an internal AI use-case register is in development
Privacy obligations apply to personal information input into a commercially available AI product, and to output that contains personal information. Published 21 October 2024 and updated 17 January 2025. It is for an organisation deploying AI that was built with, collects, stores, uses or discloses personal information, including internal use and free public chatbots. It is not a complete list of every privacy obligation.
Best practice: do not enter personal, especially sensitive, information into public generative AI tools.
A deployer includes internal use and a product or service supplied to others. The page also covers free public chatbots.
The page links three companion PDFs (top five takeaways, a selection checklist and a use checklist). They summarise this guidance.
Relevant to A25, A45, A51. Catalogue links: build a privacy-law gap table; draft a privacy-impact or algorithmic-impact assessment; build an upload checklist; decide whether client material can go into a tool. This guidance does not name those tasks.
For developers of generative AI who are APP entities and who are training or fine-tuning models, including a person who designs, builds, trains, adapts or combines models. Published 21 October 2024 and updated 23 October 2024. Publicly available data is not automatically usable to train or fine-tune a model. Developers must take reasonable steps on accuracy.
Publicly available data is not automatically usable for training.
Sensitive information generally needs consent.
Secondary use of held data for training needs careful APP 6 analysis; seek consent/opt-out if unclear.
APP 10 accuracy duties.
Relevant to A45, A25. Catalogue links: draft a privacy-impact or algorithmic-impact assessment; build a privacy-law gap table. This guidance does not name those tasks.
Review of AI use by 23 AFS and credit licensees, released 29 October 2024. It warns that governance is lagging AI adoption. It reports what ASIC found. It does not add a licence condition.
Media release 24-238MR.
Relevant to advising AFSL/ACL holders.
The review covered 624 use cases as at December 2023, and meetings with 12 of the licensees in June 2024. Only 12 of the 23 licensees had AI policy documents that referenced fairness or related concepts such as discrimination or bias.
The report says the framework is technology neutral. Licensees need to consider existing general licensee obligations, directors' duties and consumer protection provisions, including where a third party develops the model.
Relevant to A28, A3. Catalogue links: build an AI-use gap table against the licence's general obligations; draft a financial-services or credit-licence application narrative. This report does not name those tasks.
Guidance for registered tax and BAS agents on using AI while providing tax agent services, under the Code (section 30-10 of the Tax Agent Services Act 2009) and the 2024 Code Determination. Issued and last modified 22 July 2026. It does not create additional rights or legal obligations beyond the TASA.
Does not create new obligations beyond TASA.
The statement says tax practitioners remain responsible for the services they provide and must assess AI outputs with professional judgement before reliance.
Relevant to A26. Catalogue links: reconcile a tax computation; prepare an information request for a return position; build a tax due-diligence issues list. This statement does not name those tasks.
Guidance on the use of AI in delivering immigration assistanceRegulator guidanceSecondary sourceUnverified dateThis note relies on a secondary report. The primary OMARA page was not retrieved; confirm the primary text before reliance.Secondary. About immigration assistance.
Office of the Migration Agents Registration Authority
Report of the UK-Australia Legal Services Regulatory Dialogue, published by the Legal Services Council, with a section on legal technology and AI comparing regulator, professional-body and court responses in both countries.
Notes rapidly rising use of generative AI by lawyers in the UK and Australia
Summarises the NSW Law Society, LPBWA and VLSB+C joint statement that AI cannot substitute for legal knowledge and must be verified
Notes QLS-led multi-state guidance on AI selection and use
Observes common themes in Australian court practice notes (confidentiality, limits on AI in affidavits, citation verification, expert reports)
Examiner guidance. An inventor must be a natural person. A corporation or AI cannot be an inventor. The manual records Stephen L Thaler [2021] APO 5, overturned by a single judge in Thaler v Commissioner of Patents [2021] FCA 879, then restored by the Full Court in Commissioner of Patents v Thaler [2022] FCAFC 62. Special leave was refused in Thaler v Commissioner of Patents [2022] HCATrans 199 on 11 November 2022. Naming a non-human inventor fails the formalities.
The manual says the Full Court finding is the current state of the law on AI inventorship.
Relevant to A23. Catalogue links: prepare a citation list of material prior art; build a prior-art hit table; prepare an issues list on AI-generated works and training data; draft an IP licence or assignment. This manual section does not name those tasks.
Office of the Migration Agents Registration Authority
2026-03-05
OMARA guidance on how AI interacts with lawful immigration assistance, RMA responsibilities and consumer consent.
Only RMAs, Australian legal practitioners and limited exempt persons may give immigration assistance; there is no exemption for AI-generated information
A person who provides immigration assistance through AI may commit an offence under s 280 of the Migration Act
RMAs who use AI remain responsible for the accuracy of the assistance; Code of Conduct standards apply
RMAs need clients' written consent before entering personal details into an AI platform
Consumers can complain to the OMARA about an RMA's use of AI
The page was last updated on 5 March 2026.
Relevant to A30. Catalogue links: draft a client declaration or interview-question list; build a visa-pathway longlist; prepare a response-to-information-request checklist. This guidance does not name those tasks.
Member Forum Q&A - Artificial IntelligenceStatementExternal dispute schemePrimary sourceConfidence: mediumUnverified dateAFCA describes itself as an ombudsman and external dispute resolution scheme, not a statutory tribunal.Secondary. Scope not stated in the stored source.
Australian Financial Complaints Authority
2026-05-27
AFCA's answers to member questions about rising AI use by complainants and representatives and AFCA's own use of AI.
Generic or AI-generated templates with vague or unsubstantiated allegations are not appropriate
AFCA is monitoring complaint submissions, including from paid representatives
AFCA is exploring AI-enabled tools including transcription, call summaries and efficiency insights
AI does not replace human involvement, especially where decisions are made; complaints are assessed on merits and evidence
No AI guidance found as of 10 Oct 2026.No guidance foundPrimary sourceOnly submissions found (e.g. "AI: Governance and Leadership"); scrape returned no text. No practitioner guidance.
Law Council of Australia
searched
Only submissions found (e.g. "AI: Governance and Leadership"); scrape returned no text. No practitioner guidance.
OAIC publishes “New AI guidance makes privacy compliance easier for business | OAIC”. New guides for businesses published today by the Office of the Australian Information Commissioner (OAIC) clearly articulate how Australian privacy law applies to artificial intelligence (AI) and set out the regulator’s expectations. Checked 2026-10-10. Read the linked page before relying on it.
New guides for businesses published today by the Office of the Australian Information Commissioner (OAIC) clearly articulate how Australian privacy law applies to artificial intelligence (AI) and set out the regulator’s expectations.
OAIC publishes “Getting AI right benefits businesses, productivity and the community - OAIC”. Strong privacy governance and safeguards are essential for businesses to gain advantage from AI and build trust and confidence in the community. Checked 2026-10-10. Read the linked page before relying on it.
Strong privacy governance and safeguards are essential for businesses to gain advantage from AI and build trust and confidence in the community.
OAIC publishes “Can personal information be used to develop or train GenAI? - OAIC”. We cover when and why privacy should be considered by developers using personal information to train generative AI models, how our new guidance fits in with that being published globally and how we see it continuing to apply as technology changes. Checked 2026-10-10. Read the linked page before relying on it.
We cover when and why privacy should be considered by developers using personal information to train generative AI models, how our new guidance fits in with that being published globally and how we see it continuing to apply as technology changes.
This year, we have seen Australians express increasing concern about the use of their personal information to train generative artificial intelligence (AI) products. The message from the public is clear: organisations need to protect privacy.
No guidance found as of 10 Oct 2026: Australian Bar AssociationNo guidance foundBarPrimary sourceNo standalone ABA AI instrument was found as of 10 Oct 2026. The general news address is a script-rendered list and did not yield a separate ABA practice notice in this review. The Federal Court's 30 April 2025 Notice to the Profession is already recorded against the Court.
Australian Bar Association
Not stated
No standalone ABA AI instrument was found as of 10 Oct 2026. The general news address is a script-rendered list and did not yield a separate ABA practice notice in this review. The Federal Court's 30 April 2025 Notice to the Profession is already recorded against the Court.
No guidance found as of 10 Oct 2026: Copyright Tribunal of AustraliaNo guidance foundTribunalPrimary sourceNo Copyright Tribunal AI practice guidance found as of 10 Oct 2026 (homepage and practice-and-procedure page checked; both minimal). Proceedings run through the Federal Court registry, so the Federal Court GPN-AI may be relevant.
Copyright Tribunal of Australia
Not stated
No Copyright Tribunal AI practice guidance found as of 10 Oct 2026 (homepage and practice-and-procedure page checked; both minimal). Proceedings run through the Federal Court registry, so the Federal Court GPN-AI may be relevant.
The Administrative Appeals Tribunal was replaced by the Administrative Review TribunalNoticeTribunalSecondary sourceConfidence: mediumUnverified dateOn 10 October 2026 a direct request to this page did not return the document. The sentence was returned by a search retrieval of the same URL that day. No separate AAT generative-AI instrument was found. ART instruments already in the catalogue are the successor’s.Secondary. Scope not stated in the stored source.
Administrative Appeals Tribunal
2024-10-14
The Administrative Appeals Tribunal (AAT) was replaced by the Administrative Review Tribunal (the Tribunal) on 14 October 2024.
The Administrative Appeals Tribunal (AAT) was replaced by the Administrative Review Tribunal (the Tribunal) on 14 October 2024. This page contains information relating to the former AAT.
The privacy and freedom-of-information regime for Commonwealth, and what the 10 October 2026 check recorded about AI. Confirm the statute, and any commencement proclamation, before reliance.
Privacy Act 1988 (Cth) + APPs; Privacy and Other Legislation Amendment Act 2024. FOI: Freedom of Information Act 1982 (Cth). Applies to most private-sector entities over A$3m turnover and to all federal agencies; law firms handling client personal information are usually covered. The OAIC has issued guidance on commercial AI products and on training generative AI. The 2024 amendments add automated-decision transparency in privacy policies (from 10 Dec 2026) and a statutory tort for serious invasions of privacy. Checked 2026-10-10.
Privacy Act 1988 (Cth) + APPs; Privacy and Other Legislation Amendment Act 2024
FOI: Freedom of Information Act 1982 (Cth)
OAIC: Privacy and commercially available AI products — https://www.oaic.gov.au/privacy/privacy-guidance-for-organisations-and-government-agencies/guidance-on-privacy-and-the-use-of-commercially-available-ai-products
OAIC blog: training GenAI with personal information — https://www.oaic.gov.au/news/blog/can-personal-information-be-used-to-develop-or-train-a-generative-ai-model
Statute names from official sources; how they apply to AI is a summary to verify.
GPN-AI, the FCFCOA practice direction and an ART direction ask different things. Read the one that governs the file.
Verify in a real source
The Federal Court warns that generative tools can invent citations and then agree that the invention is real. Check authorities in AustLII or the Federal Register of Legislation. Do not ask a second model to bless the first.
Evidence stays personal
An affidavit must reflect the witness’s own recollection. An expert report must contain the expert’s own opinion. GPN-AI requires disclosure where the tool was used to summarise or analyse the material a witness relies on, to create media, or in a way that could affect admissibility.
Jurisdiction
This page is the Commonwealth overlay. Pair it with the practice area for the work, such as immigration, financial services, climate reporting or procurement.
What changes in Commonwealth courts
Read the note for the court on the file. The Federal Court’s list of cautions is not the family court’s list, and neither is a state Supreme Court practice note.
Review checklist
Open GPN-AI, the FCFCOA practice direction, or the ART direction, whichever governs the proceeding.
Verify citations in AustLII or the Federal Register of Legislation.
Apply the confidentiality limits in the note before any document is uploaded.
Follow the disclosure rule for affidavits, expert reports and submissions in that court.
Be ready to tell the court what tool was used and how the output was checked.
Jurisdiction-specific considerations
A state practising certificate does not make that state’s Supreme Court note the rule of the Federal Court.
Family law and migration files have extra publication limits. The FCFCOA direction treats a public upload as a possible criminal or safety problem, not only a privacy issue.
The ART direction was listed by the Law Society of New South Wales as at 14 May 2026. Read the current PDF. Do not assume it copies GPN-AI.
Further check from the 22 September research
The Administrative Review Tribunal practice direction, adapted from the Federal Court’s note, requires the user to verify assisted material and to be able to say what tool was used and how it was checked. Disclosure is required where a tool summarised or analysed the material a witness relies on, translated evidence, created media, or was used in a way that could affect how the Tribunal uses the document. Put that disclosure near the start.
Checks to carry into the matter
Identify the court or tribunal and open its current AI practice note before a draft is filed.
Verify every citation, quote and legislative reference in a source you trust. Do not use one AI tool to check another.
Keep suppressed, privileged and subpoenaed material out of a public tool.
Keep affidavits and expert reports in the witness’s or expert’s own words, and follow that court’s disclosure rule.
Apply the conduct rules of the jurisdiction where you hold a practising certificate, including competence, confidentiality and the duty to the court.
Limits on AI assistance
Do not file, send or rely on generated text until the responsible lawyer has checked it against the instructions and the current source.
Do not put suppressed, privileged, subpoenaed or safety-sensitive information into a public AI tool.
Check caps on the ATO site: from 1 July 2026 concessional $32,500, non-concessional $130,000 Check the bring-forward limit ($390,000 from 1 July 2026) Confirm TSB from the ATO portal at the right 30 June ($2M NCC cut-off in 2025-26) Check for excess notices Verify every figure
Apply RG 105 (June 2022) for competency and RG 1 (revised 16 June 2025) for process Do not use RG 2 or RG 3; they were withdrawn Lodge on the ASIC Regulatory Portal with structured answers Match RM claims to People Proofs and authorisations to each service
Check every case on AustLII Test the client's actual words and the overall impression (Campomar v Nike [2000] HCA 12) Confirm 'trade or commerce' and any ACCC orders or undertakings Check the 2026-27 drip-pricing priority
Confirm the April 2025 model disclosure document was used Check the November 2025 items: capital expenditure and specific purpose funds Check the four-month annual update and any earnings claims Re-derive every figure
Check the s 5 head, reading s 12(1)(l) narrowly Check s 11 disqualifying purposes Check PBI direct relief (Equality Australia [2024] FCAFC 115; July 2025 ACNC statement) Verify every case cited
Capital Markets & Securities — ECM/DCM and Disclosure
Trace every fact to a primary source Check ss 710–713, 715A and 716–728 on the Federal Register of Legislation Confirm forward-looking statements have reasonable grounds (s 728(1)(c)) Make risks company-specific Confirm s 713 is available Get DDC approval; note s 729 exposure
Keep the output internal until re-derived Verify every authority and the s 33C(1)(a)–(c) elements (or state equivalent) Check limitation, s 33ZE suspension and tolling Test the common questions For securities claims, apply the s 674/674A version in force at the time (pre- or post-May 2021)
Form the solvency opinion yourself (cash-flow and balance-sheet tests, s 95A) Check RG 217 (December 2024): a course of action reasonably likely to lead to a better outcome Check proper records, appropriate advice and contemporaneous documents Confirm safe harbour began before insolvency Reconcile every figure
Decide the procedure yourself Confirm determinations or claims over the land by same-day NNTT NTV search Confirm the category: expedited procedure, right to negotiate, compulsory acquisition, infrastructure Confirm s 29 notice deadlines, which vary by jurisdiction and cannot be extended Use the Native Title (ILUAs) Regulations 2024 (Cth) for new ILUAs (from 1 October 2024; 1999 Regulations replaced) Check the ALRC final report (31 March 2026): impact-based categories, 18-month RTN, five-year moratorium
Verify the facts from records Apply current legislation and ACCC-linked instruments, exclusions and exemptions Test monetary, cumulative, asset and voting-power thresholds Note: regime from 1 January 2026; asset and voting-power rules from 1 April 2026 Do not treat a checklist as clearance
Check attachment (PPSA s 19), third-party enforceability (s 20) and perfection (s 21) Check circulating status and PMSI priority Match the grantor name to the ASIC extract; s 165(b) errors defeat perfection Note secured-party errors differ (AMAL Security Services v 452HM [2025] FCA 603) For individual debtors, check the National Credit Code Check title searches
Check Schedule 6D Migration Regulations 1994 as at the EOI date Check test-specific English thresholds (IELTS/PTE/OET/TOEFL) Check the ANZSCO unit group and work periods Check the current skilled occupation list for the subclass Check recent SkillSelect rounds on immi.homeaffairs.gov.au
Check each authority on AustLII and read the paragraph Note invented cases: Valu [2025] FedCFamC2G 94 & 95, JNE24 [2025] FedCFamC2G 1314 Note also Pasuengos [2025] FedCFamC2G 2129 Quote verbatim Frame grounds under the ADJR Act 1977, s 39B Judiciary Act 1903 or constitutional writs Put the GPN-AI (16 April 2026) declaration at the top: tool, function, location
Verify every fact and citation before this Federal Court filing Remember Murray [2025] FCA 731: personal indemnity costs for false footnotes Disclose AI use as GPN-AI (16 April 2026) requires Do not enter confidential or privileged material into the tool (GPN-AI)
Verify every citation on AustLII or the FWC database (fabricated AI citations: Riley v Nuvei [2026] FWC 75) Check the s.394 time limit: 21 days from dismissal; extension only in exceptional circumstances Check the minimum employment period (s.383): 1 year for small business employers (s.23), otherwise 6 months Follow the FWC exposure draft Guidance Note (March 2026; not final or in the FWC Rules 2024) Disclose AI use, certify the content was checked and hyperlink cited cases Confirm witness statements rest on the deponent's own knowledge
Confirm independent legal advice certified under ss 90G and 90UJ FLA; a generated certificate is no substitute Verify every figure, asset, trust, company and super interest against disclosure Apply the codified four-step process (ss 79/90SM as amended) to consent orders from 10 June 2025 Include the family-violence economic factor in those orders Use super-split forms under the Family Law (Superannuation) Regulations 2025 (from 1 April 2025) Apply the 2025 Methods and Factors Approval to super valuations Check current forms on the FCFCOA eCourts Portal
Check every proposition against the current IP Australia Examination Guidelines Apply the October 2025 computer-implemented invention guidance after Aristocrat [2025] FCAFC 131 Do not apply pre-Aristocrat framing or US Alice/Mayo Check amendments do not add matter (Patents Act s 102) Verify every case, guideline and office decision exists and says what is claimed Consider what any argument or amendment concedes Do not narrow beyond commercial need
Technology, Media & Telecommunications / IT Contracts
Confirm the contract is standard form Confirm small business: under 100 staff or under AU$10M turnover; no value cap since Nov 2023 Check penalty: greatest of AU$50M, 3× benefit or 30% of adjusted turnover Check cases on AustLII Apply the ASIC Act, not the ACL, to financial services
Decide serious harm yourself; check 30 days from awareness of a suspected breach Check the s 27 Cyber Security Act 2024 ransomware report: 72 hours; turnover over $3m Note the ransomware penalty is civil, 60 penalty units Check SOCI reports to the ACSC in 12 or 72 hours Check privilege (Optus v Robertson; McClure v Medibank [2025] FCA 167) Check state health records Acts (Vic, NSW, ACT)
Check the s 14ZW TAA 1953 time limit for the tax and taxpayer Income tax: 2 years for most individuals and small and medium businesses (longer for others) GST and FBT assessments: 4 years; amended assessments: later of 60 days or the original period Check cases, grounds and the DUO reference Use the approved form (s 14ZU)
Check the instrument type: straight, order or bearer Check incorporation of charterparty terms Check the law and arbitration clause after Carmichael Rail [2024] HCA 4 Check Himalaya cover under Australian authority
Remember super follows the fund deed, not the will; verify nomination validity under this deed Check whether the deed allows non-lapsing BDBNs (many industry funds do not) Confirm nominees are 'dependants' (SIS Act s 10(1)) or the legal personal representative Check SMSF death benefit rules, which differ from APRA funds (ATO: sole purpose test, payment standards) Check reversionary pension effects on the assets test exemption and transfer balance cap Confirm the execution date of a lapsing BDBN and diarise its two-year renewal
Confirm the status of the Criminal Code Amendment (Deepfake Sexual Material) Act 2024 Confirm the status of the NSW Intimate Image and Audio Material Act 2025 and SA law Confirm the status of the My Face, My Rights Bill 2025 Check the privacy tort (10 June 2025) and OAIC biometric rulings Do not rely on a personality right: none exists
Apply the s.15AA FWA test (from 26 August 2024) from primary sources: real substance, practical reality, true nature Note s.15AA replaced the contractual-terms focus of CFMMEU v Personnel Contracting Pty Ltd [2022] HCA 1 Note it also replaced that focus in ZG Operations Australia Pty Ltd v Jamsek [2022] HCA 2 Remember a wrong call invalidates advice on awards, unfair dismissal, NES, super and payroll tax Verify the opt-out for contractors above the s.15AA threshold Check employee-like worker and regulated platform work provisions (Closing Loopholes No.2) Treat those as a separate FWC jurisdiction; confirm the WA state-system boundary (sole traders, partnerships)
Verify thresholds on foreigninvestment.gov.au on the day of advice Check agricultural land ($15M cumulative) and agribusiness ($75M in 2026) Check FTA and foreign government investor thresholds Check the 30-day register deadline and Guidance Note 7
Check deemed change-of-control triggers Check this year's FIRB thresholds Check mandatory ACCC notice under CCA Part IVA (from 1 January 2026; A$200M combined revenue) Check litigation status and that documents are the executed versions
Check the ban first: 1 April 2025 to 30 June 2029, no foreign purchases of established dwellings Check the narrow exceptions: 20+ dwelling redevelopment, commercial-scale housing, build-to-rent Get approval for new dwellings and vacant land; fees index each 1 July Check state vacant land tax and the ATO vacancy fee
Remember super sits outside the estate; the SIS Act 1993, SISR 1994 and the fund deed govern Read this fund's deed for lapsing and BDBN form; after Hill v Zuda the deed governs SMSF nominations Apply the three-year lapse and two-witness rules if the deed requires reg 6.17A compliance Check dependant status under SIS Act s 10 and, separately, ITAA 1997 s 302-195 Check tax: 0% for tax dependants; up to 17% (15% + 2% Medicare) on the taxable component otherwise Note the tax-free component stays tax-free Consider a superannuation proceeds trust; a defective BDBN leaves the trustee an unfettered discretion
Read the full award Test refusal grounds against current Federal Court authority; the s 8(7A) threshold is high Check the limitation period in the state Act (generally 6 years) Verify Convention membership and reservations on UNCITRAL Check for arbitrator AI use (LaPaglia v Valve)
Check CPS 230 (from 1 July 2025; amendments 30 April 2026; revised standard and CPG 230 from 1 July 2026) Confirm AI vendors' material service provider status against the current Attachment Note ASIC's December 2024 sweep found every reviewed insurer non-compliant with at least one RG 271 obligation Check APP 1.7 automated-decision disclosure (from 10 December 2026) and OAIC guidance (expected by September 2026) Check the current General Insurance Code on the Insurance Council of Australia site Note the revised Code (consulted Q1 2026) is expected to go to ASIC mid-2026 Apply CPS 234 to AI cyber-risk governance
Check going-concern requirements: both registered, enterprise carried on, all things necessary supplied Check the going concern and margin scheme against GSTR 2002/5, GSTR 2000/21 and any private ruling Check whether vacant possession breaks the going concern Tailor the clause and advise on GST risk in writing Consult the accountant
