On 12 November 2026 Employment Minister Amanda Rishworth released a draft Best practice guide on the adoption of AI in the workplace. AAP reported it as the first workplace AI guide from the federal government: it is about AI and staff, rather than AI and data or safety in general, and it is open for comment until 27 January 2027.
The headlines were about big employers: job cuts, offshoring and a business lobby worried about red tape. The guide itself is written for workplaces of every size and has a note for small businesses in four of its eight principles. If you employ people and you are bringing in ChatGPT, Copilot, a note taker or an AI receptionist, it describes what the government thinks good looks like. This post sets out what the draft says, what it does not change, and what a small business can sensibly do this month.
Key facts
- The Department of Employment and Workplace Relations published the draft Best practice guide on the adoption of AI in the workplace for consultation, with the minister's release on 12 November 2026.
- Consultation closes at 11:59 pm AEDT on 27 January 2027, and businesses of all sizes, workers, unions and industry groups are invited to comment.
- The draft says it is voluntary, does not create or expand legal obligations and is not a code of practice under work health and safety laws.
- It sets out 8 principles under 3 parts, Plan, Enable and Embed, running from setting goals for AI to managing risks and work health and safety.
- The draft names Microsoft Copilot, ChatGPT and Claude as among the most visible forms of AI, and says existing workplace laws apply to AI even when they do not mention it.
- The Australian Chamber of Commerce and Industry warned that consultation not proportionate to the risk and scale of a change "will slow adoption and add cost".
Quick answer: nothing in the draft is new law, so there is nothing you must do by a date. What it does is spell out an expectation: tell your staff what AI you plan to use and why, before you switch it on, and give them a real say. For a small business that can be a toolbox talk, a one-page AI policy and one named person accountable for each AI tool. Your existing consultation duties under your award, agreement and work health and safety law already apply, and the guide says so plainly.
What the draft actually says
The guide is 45 pages. Its central idea is in the introduction: best practice treats adopting high-impact AI tools "primarily as a workplace transformation, rather than a technology rollout". In plain words, the question is not which tool to buy, but what changes for the people who will use it.
The eight principles, under three parts:
| Part |
Principles |
| Plan |
1. Capturing the opportunity. 2. Identifying use cases and assessing workplace readiness. 3. Cooperative planning and worker engagement. |
| Enable |
4. Designing work and jobs. 5. Building capability. 6. Managing workforce transformation. |
| Embed |
7. Establishing AI governance. 8. Proactively managing risks and work health and safety. |
Running through all eight is trust. The draft lists factors it says build it, including transparency about how, where and why AI is used, fairness, and accountability that sits with a human decision maker rather than the system.
The guide was built from roundtables with unions, employer groups and businesses including Microsoft, Commonwealth Bank, Telstra, Kmart, Blackmores and Ramsay Health Care. It sits alongside the National AI Centre's Guidance for AI Adoption, which it points smaller organisations to for the basics.
What it does not change
The draft is careful on this, and so should we be. It says it "is voluntary in nature, does not create or expand legal obligations" and is not a code of practice under work health and safety law.
What it does do is remind employers of the obligations they already have. Workplace laws are written to be technology neutral, so they apply to AI even though they do not name it. If your modern award or enterprise agreement requires you to consult about major workplace change, that can include bringing in an AI system that changes people's work. Work health and safety law has its own consultation duties. Privacy, discrimination and consumer law still apply too. Which of these bite depends on your award, your agreement and your state, so check the Fair Work Ombudsman and Safe Work Australia pages the guide links to, or ask an adviser. This post is not legal advice.
What it asks of a small business
This is the part the coverage skipped. The draft has a small business box in four of its principles, and each one scales the expectation down rather than waiving it.
Talking to staff. Small businesses must still meet their consultation obligations, the draft says, but engagement can be informal. It suggests "toolbox talks" or similar, as long as workers have a genuine chance to shape the change, and it says engagement should start early and happen regularly.
Building skills. Formal training may be out of reach, so the draft points to on-the-job learning, mentoring, peer learning and vendor-led training. It also says small businesses "should not underestimate the power of experimentation" and suggests giving staff time to trial tools.
When jobs change. A small business has fewer roles to move people into. The draft asks for early information, notice and on-the-job training for redesigned tasks.
Who is responsible. No committee is needed, but someone must be accountable. Where AI supports a decision, the draft says the business should name the person or people responsible for the final call.
Elsewhere the draft suggests a pilot in one area before a wider rollout, and clear guidance on AI use, either in existing policies or in a dedicated AI policy written with staff.
What to do first
- Write down what you already use. List every AI tool staff use for work, approved or not, including the free accounts. Our post on choosing an AI assistant for the whole team covers how ChatGPT, Claude and Microsoft Copilot differ on business plans and data controls.
- For the next tool, say why before you buy. One sentence on the problem it solves and what it will change in someone's day. That is principle 1 and principle 2 in practice.
- Hold the toolbox talk. Fifteen minutes with the people affected: what the tool does, what it will not be used for (for example, whether it will be used to monitor performance), how to raise a problem, and when you will review it. Write down what they said.
- Run it as a pilot. One team or one job for a few weeks, then decide.
- Put it in a one-page policy and name an owner. Our guide to writing an AI use policy for a small team covers what data goes in, who approves tools, who checks the output and when you tell customers. Add one line per tool with the person accountable for it.
- Comment on the draft if you have a view. The consultation is open to businesses of all sizes until 27 January 2027.
The catch
A voluntary guide can still become the yardstick. If a dispute about AI and someone's job reaches the Fair Work Commission, or a worker raises a safety concern about an AI system, the practical question is likely to be whether the employer acted reasonably, and a published government guide is an easy place to look for what reasonable means. The small business steps are light: a talk, a page of policy and a named owner.
The other catch runs the other way. The business groups have a point: consultation can slow a change that carries little risk. The draft ties the level of assessment and engagement to the nature, size and resources of the business, and to the likely impact of the change.
Questions people ask
Is the workplace AI guide mandatory?
No. The draft says it is voluntary, does not create or expand legal obligations, and is not a code of practice under work health and safety laws. Your existing obligations under the Fair Work Act, your award or enterprise agreement and work health and safety law still apply to AI, as they do to any other workplace change.
Does the guide apply to small businesses?
Yes. The draft says it applies to workplaces of all sizes and should be adapted to each one. It includes small business notes on engaging staff, building skills, managing job changes and governance, each suggesting lighter ways to meet the same aim, such as toolbox talks instead of committees.
Do I have to consult staff before using ChatGPT or Copilot at work?
It depends on the change and on your obligations. Many modern awards and enterprise agreements require consultation about major workplace change, and work health and safety law has its own consultation duties. The draft guide asks employers to engage workers early as best practice, in proportion to the likely impact. Check the Fair Work Ombudsman's guidance for your award, or ask an adviser.
When does consultation on the draft close?
At 11:59 pm AEDT on 27 January 2027. Submissions go through the Department of Employment and Workplace Relations consultation hub, and the department says it may publish them.
How is this different from the National AI Centre's guidance?
The National AI Centre's Guidance for AI Adoption sets out six essential practices for any organisation using AI, covering safety, transparency and accountability. The new draft is narrower and workplace focused: how to involve staff, redesign jobs, build skills and handle job changes. The draft says the two are meant to be used together.
Where to start
Sources checked on 13 November 2026: the Department of Employment and Workplace Relations consultation page and the draft guide itself, read in full (the voluntary status, the eight principles, the small business sections, the closing date); reporting by AAP (the minister's announcement, the businesses consulted and the Australian Chamber of Commerce and Industry's response) and The Mandarin (the release date and the Plan, Enable and Embed structure). The minister's media release on ministers.dewr.gov.au did not respond to our automated fetch; Mirage News carries its text.