---
title: "Anti-Discrimination & Industrial Relations"
url: "https://ailawframework.com/practice/anti-discrimination-and-industrial-relations"
type: "practice_area"
site: "AI Law Framework"
id: "A15"
task_count: 25
last_reviewed: "2026-10-10"
---

# Anti-Discrimination & Industrial Relations

> Keep workplace facts, protected attributes, applicable instruments and procedural routes distinct. A lawyer evaluates competing explanations and checks deadlines, evidence and the proposed remedy.

Canonical page: https://ailawframework.com/practice/anti-discrimination-and-industrial-relations  
Last reviewed: 2026-10-10

Site index for agents: https://ailawframework.com/llms.txt

Also covers: Employment law; Fair Work; workplace relations; discrimination law; IR; EEO; adverse action; unfair dismissal; enterprise bargaining; modern awards; general protections; Respect@Work; gig economy; regulated platform work

## Key facts

- Tasks: 25
- Green: 7, Yellow: 14, Red: 4
- Levels: Green is AI drafts and the lawyer checks; Yellow is lawyer and AI draft together; Red is the lawyer writes and AI may only check, clarify or collect data.

## Principles

- **Define the matter**: Confirm the client’s instructions, jurisdiction, intended output and the information available before dividing the work.
- **Verify the authorities**: Open each relied-on authority independently. Confirm its existence, current status, proposition and any quotation or pinpoint reference.
- **Preserve the witness’s evidence**: Confirm each factual assertion with the witness and original record. Preserve the witness’s own knowledge and words, and check the court’s rules for this document.

## Checklist

- Open each relied-on authority independently. Confirm its existence, current status, proposition and any quotation or pinpoint reference.
- Confirm each factual assertion with the witness and original record. Preserve the witness’s own knowledge and words, and check the court’s rules for this document.
- Trace every relied-on figure to its source, confirm dates and assumptions, and independently recalculate totals or derived amounts.
- Calculate operative dates independently from the current rule and actual triggering event. Check exceptions and record the verified deadline.
- Check the current court or regulator requirements, prescribed form and attachments for the intended document and filing.
- Review operative clauses in the complete document, including definitions, cross-references, schedules and inconsistencies with the instructions.
- Confirm that the actual tool, feature and processing terms permit use of this information, including access, retention, training and disclosure conditions.
- Verify the applicable jurisdiction and the version of each relevant law, rule or instrument in force at the material time.
- Resolve exceptions and record the responsible lawyer’s approval of the exact output and intended use.

## Restrictions

- Confirm firm, client and applicable court or professional restrictions before using any AI feature.

## Jurisdiction note

Apply this guidance to the actual matter and jurisdiction. Confirm current requirements before relying on an example.

## Basis

Practice guidance supports planning; the responsible lawyer confirms the scope, evidence and review requirements for the matter.

Also covers: Employment law; Fair Work; workplace relations; discrimination law; IR; EEO; adverse action; unfair dismissal; enterprise bargaining; modern awards; general protections; Respect@Work; gig economy; regulated platform work

Anti-discrimination and industrial relations (IR) law governs the full employment lifecycle — hiring, terms, discipline, termination, bargaining, and tribunal litigation — under a dense, frequently amended federal and state/territory scheme. The AI risk profile is acute: strict statutory time limits (21 days for unfair dismissal), mandatory jurisdictional thresholds, a presumption-reversing burden of proof in general protections matters, rapid legislative change (Closing Loopholes Acts 2023–24; new s.15AA employee/contractor definition from 26 August 2024; Queensland Anti-Discrimination Act 1991 (Qld) amendments in legislative limbo as at mid-2026; Respect@Work positive duty enforcement from December 2023), and the Fair Work Commission (FWC) itself reporting a 70 per cent lodgement surge attributable substantially to AI-generated applications — many containing hallucinated case citations and fabricated facts. Hallucination in Australian employment law contexts is especially dangerous because AI models trained predominantly on US/UK data routinely misstate which Act applies, invent FWC Full Bench decisions, and confuse repealed provisions with current law. The FWC draft Guidance Note (exposure draft 24 March 2026; consultation closed 17 April 2026; not yet finalised as at mid-2026) and Federal Court GPN-AI (16 April 2026) have materially changed disclosure and verification obligations for all practitioners filing documents in these forums.

## Limits on AI assistance

Red lines — never in this area

## Limits on AI assistance

AI must never calculate or confirm statutory limitation periods (21-day unfair dismissal under s.394 FWA; 21-day general protections dismissal under s.366 FWA; 6-month AHRC complaint under s.46PH SDA; 6-year general protections non-dismissal) without mandatory independent lawyer verification against the current primary statutory text — a missed deadline is fatal and largely non-extendable. AI must never be used to cite a case authority in any document going to a client, tribunal or court without the lawyer independently retrieving and reading the decision on AustLII, Jade.io or an official database — hallucinated FWC decisions have been submitted to the Commission (Riley v Nuvei, January 2026) and drawn adverse comment; the Federal Court GPN-AI (16 April 2026) requires personal verification by the responsible lawyer. Identifiable witness statements, complainant details, or confidential client facts from workplace investigations must never be entered into a public or non-approved AI tool — this triggers APP 6 breach, ASCR r.9 breach, and potential waiver of legal professional privilege if the material is disclosed to the AI vendor as a third party without client consent. AI must never determine credibility findings or make factual findings in a workplace investigation — these remain exclusively the human investigator's responsibility; AI-authored credibility assessments are a procedural fairness challenge vector. AI must never determine whether a client is a national system employee or falls within a state industrial system — this is a threshold jurisdictional question, and an error means all subsequent advice (award coverage, unfair dismissal access, limitation periods, remedies) is wrong from the start. AI must never generate wage rates, penalty rate percentages, or allowance quantum for use in client advice without the lawyer verifying the current figure in the applicable FWC-approved award or FWO pay guide — rates index each July and AI training data is always at least one annual review out of date. AI must never produce witness statements or alter the content of a witness statement — the NSW Supreme Court Practice Note SC GEN 23 (effective 3 February 2025) and UCPR Amendment No.104 of 2025 specifically prohibit AI from writing or altering witness statements; witness statements must now certify that AI was not used to generate or alter their content. AI must never be presented to a client as independently verifying a legal position, and AI use in FWC documents must be disclosed — the FWC draft Guidance Note (exposure draft 24 March 2026; not yet finalised) requires disclosure of AI use, confirmation that facts are correct, and (for legal practitioners and paid agents) hyperlinks to all cited decisions; concealment is a conduct and costs risk.

- AI must never calculate or confirm statutory limitation periods (21-day unfair dismissal under s.394 FWA; 21-day general protections dismissal under s.366 FWA; 6-month AHRC complaint under s.46PH SDA; 6-year general protections non-dismissal) without mandatory independent lawyer verification against the current primary statutory text — a missed deadline is fatal and largely non-extendable.
- AI must never be used to cite a case authority in any document going to a client, tribunal or court without the lawyer independently retrieving and reading the decision on AustLII, Jade.io or an official database — hallucinated FWC decisions have been submitted to the Commission (Riley v Nuvei, January 2026) and drawn adverse comment; the Federal Court GPN-AI (16 April 2026) requires personal verification by the responsible lawyer.
- Identifiable witness statements, complainant details, or confidential client facts from workplace investigations must never be entered into a public or non-approved AI tool — this triggers APP 6 breach, ASCR r.9 breach, and potential waiver of legal professional privilege if the material is disclosed to the AI vendor as a third party without client consent.
- AI must never determine credibility findings or make factual findings in a workplace investigation — these remain exclusively the human investigator's responsibility; AI-authored credibility assessments are a procedural fairness challenge vector.
- AI must never determine whether a client is a national system employee or falls within a state industrial system — this is a threshold jurisdictional question, and an error means all subsequent advice (award coverage, unfair dismissal access, limitation periods, remedies) is wrong from the start.
- AI must never generate wage rates, penalty rate percentages, or allowance quantum for use in client advice without the lawyer verifying the current figure in the applicable FWC-approved award or FWO pay guide — rates index each July and AI training data is always at least one annual review out of date.
- AI must never produce witness statements or alter the content of a witness statement — the NSW Supreme Court Practice Note SC GEN 23 (effective 3 February 2025) and UCPR Amendment No.104 of 2025 specifically prohibit AI from writing or altering witness statements; witness statements must now certify that AI was not used to generate or alter their content.
- AI must never be presented to a client as independently verifying a legal position, and AI use in FWC documents must be disclosed — the FWC draft Guidance Note (exposure draft 24 March 2026; not yet finalised) requires disclosure of AI use, confirmation that facts are correct, and (for legal practitioners and paid agents) hyperlinks to all cited decisions; concealment is a conduct and costs risk.

## Jurisdiction-specific considerations

Australian traps

## Jurisdiction-specific considerations

FWC draft Guidance Note on Generative AI (exposure draft 24 March 2026; consultation closed 17 April 2026; not yet finalised as at mid-2026): when finalised, will require any party using AI to prepare a lodged document to (a) disclose AI use, (b) confirm all facts are correct and relevant, and (c) for legal practitioners and paid agents, include hyperlinks to all cited case law. Non-compliance risks dismissal of application or costs orders. The note is not yet in force but FWC members are already adverse to AI-generated hallucinations and unverified citations. Monitor the FWC website for finalisation. Fair Work Act 2009 (Cth) s.394 — 21-day unfair dismissal time limit is absolute subject to a narrow 'exceptional circumstances' discretion. AI miscalculation of this date has already caused claims to be struck out. The trigger is the date dismissal takes effect, not the date of notification — AI frequently uses the wrong trigger event. Never rely on AI date arithmetic. High income threshold — current figure (1 July 2025 to 30 June 2025): $183,100 (reg 2.13 Fair Work Regulations 2009 (Cth)); the 2024–25 figure of $175,000 is the figure AI most commonly produces. The threshold is indexed on 1 July each year — verify the current figure in the FWC's published determination before every matter. Section 15AA FWA — new employee/employer definition (from 26 August 2024): the Closing Loopholes No.2 Act replaced the purely contractual characterisation test from CFMMEU v Personnel Contracting [2022] HCA 1 and Jamsek [2022] HCA 2 with a 'real substance, practical reality and true nature' whole-of-relationship test. AI models trained before late 2024 apply the pre-August 2024 contractual test, which is now wrong for most purposes. An opt-out mechanism exists for contractors above the designated high-income threshold. Queensland Anti-Discrimination Act 1991 (Qld) amendments: the Respect at Work and Other Matters Amendment Act 2024 (Qld) was passed September 2024; representative complaints commenced 14 October 2024; the remaining provisions (new protected attributes, positive duty, other reforms) were scheduled to commence 1 July 2025 but the Crisafulli Government announced a delay pending stakeholder consultation, with a new commencement date to be set by proclamation. As at mid-2026 that date has not been set. AI trained before the delay announcement will state these provisions commenced 1 July 2025 — this is wrong. Respect@Work positive duty — Sex Discrimination Act 1984 (Cth) s.47C: AHRC has active enforcement powers from 12 December 2023 (compliance notices, enforceable undertakings, Federal Court referral). The Work Health and Safety (Sexual and Gender-based Harassment) Code of Practice 2025 (Cth) (approved 5 March 2025) and the Work Health and Safety (Managing Psychosocial Hazards at Work) Code of Practice 2024 (Cth) add parallel WHS obligations. AI trained before mid-2025 may not reflect the 2025 Code. Right to disconnect — s.333M FWA — commenced 26 August 2024 (employers other than small businesses) and 26 August 2025 (small businesses with fewer than 15 employees). AI models trained before August 2024 have no knowledge of this provision. Any advice on out-of-hours contact obligations must verify against the current statutory text. Federal Court GPN-AI Practice Note (16 April 2026): requires the responsible lawyer to personally verify before filing that (a) facts are properly provable, (b) cited authorities exist and support the propositions advanced, (c) evidence is admissible, and (d) discovery obligations are met. Applies to all persons filing in the Federal Court. State court equivalents include NSW Supreme Court Practice Note SC GEN 23 (3 February 2025), District Court of NSW General Practice Note 2 (February 2025), and NCAT Procedural Direction No.7 (April 2025). The Law Society of NSW AI hub maintains a current court-by-court register (last updated 14 May 2026) — consult before any filing. Closing Loopholes Acts No.1 (2023) and No.2 (2024) made extensive amendments to FWA: right to disconnect (s.333M), regulated labour hire, casual conversion (new employee-choice pathway replacing prior conversion mechanism), new 'employee-like worker' and regulated platform work provisions, new definition of 'employee' (s.15AA), intractable bargaining regime, and BOOT changes for single-enterprise agreements from 27 February 2024. AI trained before late 2024 will reflect pre-Closing-Loopholes law on multiple IR questions. FWO Annual Wage Review: minimum wages and award rates update each 1 July. AI-generated wage figures are almost certainly stale for the current financial year. Always verify against the Fair Work Ombudsman's current pay guide or the FWC wage-fixing determination — do not use AI-stated figures in any client advice, settlement calculation or compliance audit. Deloitte hallucination precedent (Australia, October 2025): a $440,000 Deloitte report to the Department of Employment and Workplace Relations (published July 2025) contained fabricated academic references, a fabricated quote attributed to a Federal Court judgment (with a misspelled judge's name), and references to non-existent studies. Deloitte issued a partial refund after a Sydney University researcher publicly identified the errors in October 2025. This is a directly on-point Australian precedent for professional liability exposure from unverified AI output in the employment law and government-advice context. National AI Plan (December 2025): Australia confirmed it will not introduce a standalone AI Act; oversight sits with existing sector regulators (AHRC, FWO, OAIC, ACCC) guided by voluntary frameworks and the new AI Safety Institute. For client advice on AI-in-HR governance, there is no single AI compliance checklist — obligations are distributed across FWA, Privacy Act, anti-discrimination statutes, WHS law and Australian Consumer Law. Multi-Act due diligence is required.

- FWC draft Guidance Note on Generative AI (exposure draft 24 March 2026; consultation closed 17 April 2026; not yet finalised as at mid-2026): when finalised, will require any party using AI to prepare a lodged document to (a) disclose AI use, (b) confirm all facts are correct and relevant, and (c) for legal practitioners and paid agents, include hyperlinks to all cited case law. Non-compliance risks dismissal of application or costs orders. The note is not yet in force but FWC members are already adverse to AI-generated hallucinations and unverified citations. Monitor the FWC website for finalisation.
- Fair Work Act 2009 (Cth) s.394 — 21-day unfair dismissal time limit is absolute subject to a narrow 'exceptional circumstances' discretion. AI miscalculation of this date has already caused claims to be struck out. The trigger is the date dismissal takes effect, not the date of notification — AI frequently uses the wrong trigger event. Never rely on AI date arithmetic.
- High income threshold — current figure (1 July 2025 to 30 June 2025): $183,100 (reg 2.13 Fair Work Regulations 2009 (Cth)); the 2024–25 figure of $175,000 is the figure AI most commonly produces. The threshold is indexed on 1 July each year — verify the current figure in the FWC's published determination before every matter.
- Section 15AA FWA — new employee/employer definition (from 26 August 2024): the Closing Loopholes No.2 Act replaced the purely contractual characterisation test from CFMMEU v Personnel Contracting [2022] HCA 1 and Jamsek [2022] HCA 2 with a 'real substance, practical reality and true nature' whole-of-relationship test. AI models trained before late 2024 apply the pre-August 2024 contractual test, which is now wrong for most purposes. An opt-out mechanism exists for contractors above the designated high-income threshold.
- Queensland Anti-Discrimination Act 1991 (Qld) amendments: the Respect at Work and Other Matters Amendment Act 2024 (Qld) was passed September 2024; representative complaints commenced 14 October 2024; the remaining provisions (new protected attributes, positive duty, other reforms) were scheduled to commence 1 July 2025 but the Crisafulli Government announced a delay pending stakeholder consultation, with a new commencement date to be set by proclamation. As at mid-2026 that date has not been set. AI trained before the delay announcement will state these provisions commenced 1 July 2025 — this is wrong.
- Respect@Work positive duty — Sex Discrimination Act 1984 (Cth) s.47C: AHRC has active enforcement powers from 12 December 2023 (compliance notices, enforceable undertakings, Federal Court referral). The Work Health and Safety (Sexual and Gender-based Harassment) Code of Practice 2025 (Cth) (approved 5 March 2025) and the Work Health and Safety (Managing Psychosocial Hazards at Work) Code of Practice 2024 (Cth) add parallel WHS obligations. AI trained before mid-2025 may not reflect the 2025 Code.
- Right to disconnect — s.333M FWA — commenced 26 August 2024 (employers other than small businesses) and 26 August 2025 (small businesses with fewer than 15 employees). AI models trained before August 2024 have no knowledge of this provision. Any advice on out-of-hours contact obligations must verify against the current statutory text.
- Federal Court GPN-AI Practice Note (16 April 2026): requires the responsible lawyer to personally verify before filing that (a) facts are properly provable, (b) cited authorities exist and support the propositions advanced, (c) evidence is admissible, and (d) discovery obligations are met. Applies to all persons filing in the Federal Court. State court equivalents include NSW Supreme Court Practice Note SC GEN 23 (3 February 2025), District Court of NSW General Practice Note 2 (February 2025), and NCAT Procedural Direction No.7 (April 2025). The Law Society of NSW AI hub maintains a current court-by-court register (last updated 14 May 2026) — consult before any filing.
- Closing Loopholes Acts No.1 (2023) and No.2 (2024) made extensive amendments to FWA: right to disconnect (s.333M), regulated labour hire, casual conversion (new employee-choice pathway replacing prior conversion mechanism), new 'employee-like worker' and regulated platform work provisions, new definition of 'employee' (s.15AA), intractable bargaining regime, and BOOT changes for single-enterprise agreements from 27 February 2024. AI trained before late 2024 will reflect pre-Closing-Loopholes law on multiple IR questions.
- FWO Annual Wage Review: minimum wages and award rates update each 1 July. AI-generated wage figures are almost certainly stale for the current financial year. Always verify against the Fair Work Ombudsman's current pay guide or the FWC wage-fixing determination — do not use AI-stated figures in any client advice, settlement calculation or compliance audit.
- Deloitte hallucination precedent (Australia, October 2025): a $440,000 Deloitte report to the Department of Employment and Workplace Relations (published July 2025) contained fabricated academic references, a fabricated quote attributed to a Federal Court judgment (with a misspelled judge's name), and references to non-existent studies. Deloitte issued a partial refund after a Sydney University researcher publicly identified the errors in October 2025. This is a directly on-point Australian precedent for professional liability exposure from unverified AI output in the employment law and government-advice context.
- National AI Plan (December 2025): Australia confirmed it will not introduce a standalone AI Act; oversight sits with existing sector regulators (AHRC, FWO, OAIC, ACCC) guided by voluntary frameworks and the new AI Safety Institute. For client advice on AI-in-HR governance, there is no single AI compliance checklist — obligations are distributed across FWA, Privacy Act, anti-discrimination statutes, WHS law and Australian Consumer Law. Multi-Act due diligence is required.

## Risks to address

How it goes wrong

## Risks to address

Hallucinated FWC case citations: AI invents plausible-sounding Full Bench or Commission decisions that do not exist. Confirmed Australian example: Riley v Nuvei Australia Merchant Services (FWC, January 2026) — the Commission noted AI-generated citations 'may have been AI hallucinations and lacked any legal basis'. The FWC's draft Guidance Note hyperlink requirement (once finalised) will expose hallucinated citations mechanically. Stale high income threshold: AI states $175,000 (the 2024–25 figure) when the current threshold from 1 July 2025 is $183,100 — causing incorrect advice on whether a client can access unfair dismissal, and incorrect compensation cap calculations. Wrong 21-day unfair dismissal date: AI miscalculates from the wrong trigger event (date of notification vs date dismissal takes effect) or applies a US/UK limitation period, causing the application to be filed out of time and dismissed — effectively destroying the client's cause of action. Application of pre-August 2024 contractor test: AI applies the purely contractual characterisation test from Personnel Contracting/Jamsek (overturned for FWA purposes by s.15AA from 26 August 2024) and incorrectly advises that a contractor cannot access FWA entitlements, or vice versa. Qld ADA commenced-in-error advice: AI states that Queensland's new protected attributes (potential pregnancy, appearance, family/carer responsibilities, etc.) and positive duty commenced 1 July 2025 — they did not; commencement remains unset by proclamation as at mid-2026. Advice based on this error is wrong as to the applicable law. Stale award/wage rate: AI states a minimum wage, penalty rate, or classification rate from a prior FWC Annual Wage Review, leading to incorrect underpayment/backpay calculations that expose the client to FWO enforcement action or the firm to negligence liability.

- Hallucinated FWC case citations: AI invents plausible-sounding Full Bench or Commission decisions that do not exist. Confirmed Australian example: Riley v Nuvei Australia Merchant Services (FWC, January 2026) — the Commission noted AI-generated citations 'may have been AI hallucinations and lacked any legal basis'. The FWC's draft Guidance Note hyperlink requirement (once finalised) will expose hallucinated citations mechanically.
- Stale high income threshold: AI states $175,000 (the 2024–25 figure) when the current threshold from 1 July 2025 is $183,100 — causing incorrect advice on whether a client can access unfair dismissal, and incorrect compensation cap calculations.
- Wrong 21-day unfair dismissal date: AI miscalculates from the wrong trigger event (date of notification vs date dismissal takes effect) or applies a US/UK limitation period, causing the application to be filed out of time and dismissed — effectively destroying the client's cause of action.
- Application of pre-August 2024 contractor test: AI applies the purely contractual characterisation test from Personnel Contracting/Jamsek (overturned for FWA purposes by s.15AA from 26 August 2024) and incorrectly advises that a contractor cannot access FWA entitlements, or vice versa.
- Qld ADA commenced-in-error advice: AI states that Queensland's new protected attributes (potential pregnancy, appearance, family/carer responsibilities, etc.) and positive duty commenced 1 July 2025 — they did not; commencement remains unset by proclamation as at mid-2026. Advice based on this error is wrong as to the applicable law.
- Stale award/wage rate: AI states a minimum wage, penalty rate, or classification rate from a prior FWC Annual Wage Review, leading to incorrect underpayment/backpay calculations that expose the client to FWO enforcement action or the firm to negligence liability.

## Review checklist

The lawyer's review checklist

## Review checklist

Verify every statutory time limit cited in the AI output against the current FWA or applicable Act text — do not trust AI-calculated dates under any circumstances; confirm the correct trigger event (e.g., date dismissal takes effect, not notification date). Independently retrieve and read every case citation on AustLII or Jade.io before including in any document — confirm citation format, paragraph numbers cited, and that the decision has not been overturned or distinguishable; this is now a GPN-AI obligation for Federal Court filings. Confirm the applicable legislative scheme (federal vs state; national system vs WA state system; which anti-discrimination Act; s.15AA employee vs contractor) for the specific client, jurisdiction and employment decision type — do not accept AI's characterisation without primary source verification. Check all wage rates, penalty rates, allowances and the high income threshold ($183,100 from 1 July 2025; indexed annually) against the current FWC-approved instrument or FWO Pay Calculator — AI figures are stale and frequently cite the prior year's figures. Confirm no confidential client information (complainant identity, witness statements, dismissal facts, negotiating positions, salary details) has been entered into an unapproved AI tool — ask the fee earner specifically before relying on any AI-generated employment law document. Verify that the most recent legislative amendments are reflected in the AI output: s.15AA whole-of-relationship employee test (from 26 August 2024); s.333M right to disconnect (26 August 2024 / 26 August 2025); Closing Loopholes casual conversion, regulated labour hire and intractable bargaining provisions; Respect@Work positive duty enforcement (from 12 December 2023); WHS Sexual and Gender-based Harassment Code of Practice 2025; Queensland ADA delay (commencement date still unset). For any document to be filed with the FWC: comply with the draft Guidance Note disclosure and verification requirements (disclosure of AI use; confirmation of factual accuracy; hyperlinks to cited decisions for legal practitioners and paid agents) — even though the Guidance Note is not yet finalised, FWC members are already adverse to unverified AI content. Where AI output characterises indirect discrimination, adverse action causation, the reverse onus under s.361 FWA, or the BOOT for enterprise agreements, have the substantive legal analysis reviewed by the supervising lawyer before any advice is sent — these involve nuanced Australian doctrinal positions that AI routinely misstates, including by importing US/UK doctrine.

- Verify every statutory time limit cited in the AI output against the current FWA or applicable Act text — do not trust AI-calculated dates under any circumstances; confirm the correct trigger event (e.g., date dismissal takes effect, not notification date).
- Independently retrieve and read every case citation on AustLII or Jade.io before including in any document — confirm citation format, paragraph numbers cited, and that the decision has not been overturned or distinguishable; this is now a GPN-AI obligation for Federal Court filings.
- Confirm the applicable legislative scheme (federal vs state; national system vs WA state system; which anti-discrimination Act; s.15AA employee vs contractor) for the specific client, jurisdiction and employment decision type — do not accept AI's characterisation without primary source verification.
- Check all wage rates, penalty rates, allowances and the high income threshold ($183,100 from 1 July 2025; indexed annually) against the current FWC-approved instrument or FWO Pay Calculator — AI figures are stale and frequently cite the prior year's figures.
- Confirm no confidential client information (complainant identity, witness statements, dismissal facts, negotiating positions, salary details) has been entered into an unapproved AI tool — ask the fee earner specifically before relying on any AI-generated employment law document.
- Verify that the most recent legislative amendments are reflected in the AI output: s.15AA whole-of-relationship employee test (from 26 August 2024); s.333M right to disconnect (26 August 2024 / 26 August 2025); Closing Loopholes casual conversion, regulated labour hire and intractable bargaining provisions; Respect@Work positive duty enforcement (from 12 December 2023); WHS Sexual and Gender-based Harassment Code of Practice 2025; Queensland ADA delay (commencement date still unset).
- For any document to be filed with the FWC: comply with the draft Guidance Note disclosure and verification requirements (disclosure of AI use; confirmation of factual accuracy; hyperlinks to cited decisions for legal practitioners and paid agents) — even though the Guidance Note is not yet finalised, FWC members are already adverse to unverified AI content.
- Where AI output characterises indirect discrimination, adverse action causation, the reverse onus under s.361 FWA, or the BOOT for enterprise agreements, have the substantive legal analysis reviewed by the supervising lawyer before any advice is sent — these involve nuanced Australian doctrinal positions that AI routinely misstates, including by importing US/UK doctrine.

## Tasks (25)

| Task | Level | How AI is used | What the lawyer must review |
| --- | --- | --- | --- |
| Draft a discrimination complaint and the respondent's response | Yellow | From the chronology and documents, draft a complaint or response mapped to the alleged protected attribute, detriment and comparator, with a document list. | Identify the correct forum and time limit Do not file a generated complaint Check health or other sensitive records are handled under the applicable privacy… |
| [Build an employee-versus-contractor classification issues table](https://ailawframework.com/practice/anti-discrimination-and-industrial-relations/build-an-employee-versus-contractor-classification-issues-table.md) | Red | From the contracts, rosters and how work is actually done, table control, integration, delegation and payment facts bearing on employee versus contractor… | Decide classification on the whole relationship; a wrong call invalidates later entitlements advice Confirm which test currently applies |
| Map forum, standing and time limits for the discrimination or workplace claim | Red | From the facts brief, table candidate forums, standing rules and limitation/time-limit risks with a list of facts still needed. | Decide the forum yourself Verify the current time limit in the statute that actually applies |
| [Assess discrimination risk in the client's recruitment or performance AI tools](https://ailawframework.com/practice/anti-discrimination-and-industrial-relations/assess-discrimination-risk-in-the-client-s-recruitment-or-performance-ai.md) | Yellow | From the vendor description and the client's process, list protected-attribute risk points, documentation gaps and questions for the vendor. | Treat this as advice on the client's system, not a software assessment of the firm's tools Decide the risk yourself; the list is not a clearance |
| [Draft conciliation submissions and a deed of release](https://ailawframework.com/practice/anti-discrimination-and-industrial-relations/draft-conciliation-submissions-and-a-deed-of-release.md) | Yellow | Draft a deed of release and conciliation terms from the agreed heads, flagging tax, retirement contributions, confidentiality and regulator-reporting gaps. | Decide the release scope yourself Confirm tax treatment of termination payments independently Check confidentiality does not block lawful regulator reports |
| Build a discrimination chronology and comparator table from the file | Yellow | From permitted emails, rosters and policies, draft a dated chronology and a comparator table (treatment of the complainant vs named comparators) with document… | Decide who is a true comparator; it is a legal issue Do not invent comparators |
| Compare a workplace policy to the applicable discrimination duties | Yellow | Map the client's policy against the lawyer-selected duty checklist; list gaps, conflicts with other policies, and practices on file that do not match the… | Build the duty checklist from the current statute and any binding standard Check actual practice; policy language is not compliance if practice differs |
| Draft a reasonable-adjustments issues list from the medical and role documents | Yellow | From the role description, medical certificates the lawyer has cleared, and the request, list possible adjustments, operational impacts and information still… | Decide unjustifiable hardship or reasonableness yourself Do not load health records into a tool without a clearance decision |
| Map the client's measures against the positive-duty standards | Yellow | Read the client's policies, training records, complaint and reporting data, risk materials and board papers, and draft a table mapping each existing measure… | Confirm which positive duty applies to this entity and jurisdiction, verify each mapped measure against the source document, and decide what is reasonable and… |
| Prepare the protected action ballot application pack | Yellow | Draft the Form F34 fields and the draft order to the Commission's requirements from the lawyer's instructions on who is balloted and what action is… | Settle each ballot question against the forms of action actually contemplated and the statutory limits, confirm the nominal expiry date and that lodgement is… |
| Check a right-of-entry notice and prepare the site response | Green | Check the entry notice and permit details against the stated requirements for the entry purpose, and draft the compliance note and the site instruction sheet… | Confirm the permit is current and not suspended and that any exemption certificate is genuine, decide whether the entry is under the Fair Work Act or the work… |
| Issue the notice of employee representational rights and open the bargaining record | Green | Populate the prescribed notice with the employer's legal name, the proposed agreement name and the proposed coverage, and build the issue-and-service record… | Fix the notification time, compare the notice against the prescribed form word for word so that nothing has been added, removed or reformatted, confirm the… |
| Build the pre-approval steps record for an enterprise agreement approval | Red | Extract the date, method and recipients of each pre-approval step from the bargaining documents into a source-referenced record, compare the dates against the… | Verify every date against the primary document rather than the record, decide whether the explanation given to employees was appropriate for that workforce,… |
| Take a discrimination complaint from the commission to the court or tribunal | Yellow | Draft the application or referral response for the chosen forum from the lawyer's settled grounds, carry the complaint's particulars and the respondent… | Independently calculate the deadline from the date on the notice and decide whether an extension application is needed, settle the grounds of unlawful… |
| Prepare a work-related sexual harassment application and record the forum election | Yellow | Set out the available routes with their time limits and what each one forecloses, and draft the application fields and the outcomes sought from the lawyer's… | Confirm the date the conduct occurred or last occurred against the commencement boundary so the correct form is used, confirm eligibility and the two-year… |
| Build the annual compliance and lodgement pack for a registered organisation | Green | Draft the timetable and the disclosure checklist for each reporting unit from the required disclosures in the applicable reporting guidelines and the… | Confirm the reporting units and whether any certificate dividing them differently, or any income-based certificate or exemption, is in force for the year,… |
| Build the adjustments and consultation record for a student with disability | Green | Extract the consultation and adjustment events from the provider's files — enrolment records, meeting notes, learning plans, correspondence — into a dated… | Confirm each entry against the source document, decide whether the adjustment was reasonable and whether consultation with the student or associate actually… |
| [Audit an AI recruitment tool for discrimination risk](https://ailawframework.com/practice/anti-discrimination-and-industrial-relations/audit-an-ai-recruitment-tool-for-discrimination-risk.md) | Yellow | Map the tool's inputs and outputs against protected attributes. | Lawyer assesses direct/indirect discrimination under state and federal Acts. |
| [Advise an employer on human review of AI HR decisions](https://ailawframework.com/practice/anti-discrimination-and-industrial-relations/advise-an-employer-on-human-review-of-ai-hr-decisions.md) | Yellow | Draft a review-step procedure for AI-assisted recruitment and performance decisions. | Confirm against discrimination and Fair Work obligations. |
| [Check a self-represented applicant's AI-drafted discrimination complaint](https://ailawframework.com/practice/anti-discrimination-and-industrial-relations/check-a-self-represented-applicant-s-ai-drafted-discrimination-complaint.md) | Red | Compare complaint with commission requirements and flag invented law. | Lawyer verifies every cited authority. |
| [Prepare a Fair Work Commission AI-use disclosure for a filing](https://ailawframework.com/practice/anti-discrimination-and-industrial-relations/prepare-a-fair-work-commission-ai-use-disclosure-for-a-filing.md) | Green | Draft the AI-use statement required from 20 October 2026. | Check the final FWC guidance note wording. |
| [Summarise an employer's AI-related policy suite against workplace law](https://ailawframework.com/practice/anti-discrimination-and-industrial-relations/summarise-an-employer-s-ai-related-policy-suite-against-workplace-law.md) | Yellow | Tabulate AI, surveillance and privacy policies against obligations. | Lawyer confirms state surveillance law applies. |
| Assess an employer's use of AI in hiring for discrimination risk | Yellow | Map the AI screening steps against discrimination law. | Lawyer gives the advice. |
| Prepare a guide for a self-represented complainant using AI | Green | Explain verification and disclosure duties when using AI. | Lawyer approves. |
| Summarise a complaint and the respondent's response into an issues table | Green | Tabulate allegations, responses and evidence. | Verify against the documents. |

## Jurisdiction notes

### Australia

#### Scope

Hiring, terms, discipline, termination, bargaining and tribunal litigation, under a dense, frequently amended federal and state/territory scheme.

#### Deadlines and onus

Strict time limits (21 days for unfair dismissal), jurisdictional thresholds, and a reversed burden of proof in general protections matters.

#### Recent change

Closing Loopholes Acts 2023–24; s.15AA employee/contractor definition from 26 August 2024; Respect@Work positive duty enforced from December 2023; Anti-Discrimination Act 1991 (Qld) amendments unsettled as at mid-2026.

#### Court AI rules

FWC draft Guidance Note (exposure draft 24 March 2026; comments closed 10 April 2026; not final as at mid-2026) and Federal Court GPN-AI (16 April 2026) change disclosure and verification duties.

#### Citations

The FWC reports a surge in AI-generated applications, many with fabricated citations and facts. Check which Act applies and that every FWC Full Bench decision cited exists.

### New Zealand

Local name: Anti-discrimination and industrial relations (Human Rights Act 1993 / ERA)

#### Discrimination

Human Rights Act 1993, not the Fair Work Act general protections or the AHRC.

#### Personal grievances

Unjustified dismissal and disadvantage stay under the Employment Relations Act 2000, in the Employment Relations Authority.

#### Tribunal

The Human Rights Review Tribunal is listed in the Courts of New Zealand GenAI guidelines.

Sources:

- [Human Rights Act 1993](https://www.legislation.govt.nz/act/public/1993/0082/latest/dlm304212.html) (checked 2026-09-23)
- [Employment Relations Act 2000](https://www.legislation.govt.nz/act/public/2000/24/en/latest/) (checked 2026-09-23)
- [Employment New Zealand: employment legislation](https://www.employment.govt.nz/employment-new-zealand/employment-legislation) (checked 2026-09-23)
- [Courts of New Zealand: GenAI guidelines — Lawyers (7 December 2023)](https://www.courtsofnz.govt.nz/assets/6-Going-to-Court/practice-directions/practice-guidelines/all-benches/20231207-GenAI-Guidelines-Lawyers.pdf) (checked 2026-09-23)

### United States

Local name: Anti-discrimination and labour relations

#### Governing law

Title VII, ADA, ADEA, Equal Pay Act, GINA and the National Labor Relations Act, plus state fair-employment and labour statutes.

#### AI tools

The EEOC April 2024 explainer treats AI and automated systems as ordinary employment practices for discrimination analysis.

#### Labour relations

Union and unfair-labour-practice work is the NLRB and state agencies, not the Fair Work Commission.

Sources:

- [EEOC, What is the EEOC’s role in AI? (29 April 2024)](https://www.eeoc.gov/sites/default/files/2024-04/20240429_What%20is%20the%20EEOCs%20role%20in%20AI.pdf) (checked 2026-09-23)
- [EEOC: Artificial Intelligence and the ADA](https://www.eeoc.gov/eeoc-disability-related-resources/artificial-intelligence-and-ada) (checked 2026-09-23)
- [ABA Formal Opinion 512, Generative Artificial Intelligence Tools (29 July 2024)](https://www.americanbar.org/content/dam/aba/administrative/professional_responsibility/ethics-opinions/aba-formal-opinion-512.pdf) (checked 2026-09-23)
- [EEOC: Age Discrimination in Employment Act of 1967](https://www.eeoc.gov/statutes/age-discrimination-employment-act-1967) (checked 2026-09-23)
- [EEOC: Equal Pay Act of 1963](https://www.eeoc.gov/statutes/equal-pay-act-1963) (checked 2026-09-23)
- [EEOC: Genetic Information Nondiscrimination Act of 2008](https://www.eeoc.gov/statutes/genetic-information-nondiscrimination-act-2008) (checked 2026-09-23)
- [29 U.S.C. § 151 (National Labor Relations Act)](https://www.law.cornell.edu/uscode/text/29/151) (checked 2026-09-23)

*Practice guidance for planning and review. Confirm current primary sources and the requirements of the actual matter. This is general information, not legal advice.*