The federal government is building a national AI framework. Here's what it means for Australian small and medium sized business, and why support beats more red tape.
The federal government has decided AI needs a single plan. In a speech in Sydney, Prime Minister Anthony Albanese announced a new national framework for artificial intelligence and a new Office of AI to sit inside the Department of the Prime Minister and Cabinet. Its job is to pull together the work already spread across government and to coordinate a set of new “Australian Standards” for the technology. The government says this ends its old “issue-by-issue” approach and replaces it with one coordinated response covering data centres, copyright, jobs, and more.
If you run a small or medium sized business, the first thing to know is that this probably won’t put a new form on your desk tomorrow. Most small firms use AI through everyday tools, and the framework is aimed higher than that, at data centres, copyright, defence, and the standards that vendors will have to meet. The real question sits underneath the announcement. Will these new rules help you adopt AI, or will they add another layer you have to answer to before you’ve even started?
That question matters because staying competitive now depends on using AI well and cost effectively, and the burden of getting the rules wrong falls hardest on the businesses with the least time to spare.
What the framework actually does
Strip away the announcement language and the plan is mostly about coordination. Right now the states pull in different directions, especially on the data centres that AI runs on. New South Wales treats large centres as “state significant development”. Victoria usually sends big proposals through its planning minister. South Australia released a strategy in June that ties government support to sorted-out energy supply and sustainable water use. Western Australia has no special category at all. For an investor deciding where to build, that patchwork is hard to read, and the government’s pitch is that one national framework fixes it. Mr Albanese says getting this right will “enhance our appeal to international investors by delivering greater clarity and speed for approvals”.
The Office of AI isn’t starting from scratch either. The plan is to plug it into the watchdogs that already exist, like the Office of the Australian Information Commissioner, rather than build a brand new rulebook from nothing. On that reading, the framework is less a new cage and more an attempt to tidy a mess the government admits it made by handling AI piece by piece. Mr Albanese even points to past technologies, from civil aviation in the 1920s to genetics in the 1990s, as cases where a coordinated approach was the sensible move.
There’s a fair version of this plan, and it’s worth saying so plainly before picking at the risks.
Where the worry is fair
The risk isn’t the goal. It’s the shape. A framework meant to give clarity can just as easily grow into a separate agency with its own forms, its own audits, and its own idea of what counts as acceptable AI. That’s the part small and medium sized businesses should watch.
Take a small accounting firm in South East Queensland. They handle sensitive client information every day, so they already work under the Privacy Act 1988 and the Australian Privacy Principles. They want to use AI to speed up the slow parts of their week, the sorting and first-pass review that eats a junior’s Monday and Tuesday, exactly the sort of work that AI strategy for professional services is meant to sort out. What they don’t want is to find that using it means a second compliance regime sitting on top of the privacy rules they already follow, with a new body deciding after the fact whether their choices were sound.
There’s also a fair question about whether government is the right judge of what’s acceptable in a field moving this fast. The people writing the standards are rarely the people building with the tools, and a rule that made sense the day it was written can be out of date within a year (or even a month at the rate the industry is moving currently). When that happens, the cost isn’t only paperwork. It’s the investment that quietly goes elsewhere, to places where a business can move faster with less to prove first.
None of this is an argument against protecting people. Consumers and financial products need real guardrails, and AI can cause real harm when it’s used carelessly. The point is narrower. Australia already has consumer protection, privacy, and financial rules that work. Updating those to cover AI plainly would give people the safety they deserve without asking every business to learn a separate system. A new agency that duplicates what’s already there protects no one better. It just adds a door everyone has to knock on.
Support matters more than rules
Here’s the part the framework can’t do on its own. Rules tell you what you can’t do. They don’t help you do the thing. And adoption, not caution, is where most Australian businesses are actually stuck.
The numbers tell the story. Government data shows only just more than a third of small and medium businesses have adopted AI, but most use it only for internal jobs like drafting content and basic analysis, not for the work that reaches customers. The National AI Centre has found that 65% of the small and medium businesses still holding back do so for one reason, they don’t trust AI to make decisions. What’s missing there isn’t another warning. It’s help needed to start with AI in a safe and effective manner.
That’s where clear rules and real support have to travel together. Our accounting firm from before doesn’t need another lecture on risk. They need someone to walk them through an AI strategy built around their firm, show them which slow job to point AI at first, keep a person in the loop on anything that touches a client, and prove they did. Get that right and the same standards the government is drafting become easy to meet, because good practice was built in from day one. A few weeks in, a firm like that might be letting AI clear the routine 70%-80% of its document sorting while the team keeps the judgement calls.
What this means for you
The framework is coming, and on balance the coordination behind it is not a bad thing. Clarity for investors is real, and a single playbook beats a patchwork. The danger is only if clarity turns into a separate bureaucracy that treats every small and medium sized business like a frontier risk. That part is worth pushing back on, and the better path is to fold AI into the consumer protections Australia already has.
But you don’t have to wait for any of that to be settled. The businesses that come through this well will be the ones already using AI carefully by the time the standards land, not the ones frozen in place waiting to be told it’s allowed. If you want to be in the first group, take our AI Roadmap Interview. You’ll talk through your team’s goals and pain points, and get a custom plan for where to start and how to stay on the right side of the rules from day one.
