AI Compliance & Regulation
Does Australia's New AI Safety Regulator Apply to Your Business?
Canberra will legislate mandatory AI safety rules and a new regulator for frontier AI firms. What it means for Australian SMEs using AI tools and what to do.
In short. Not directly. The mandatory AI safety rules announced on 10 November 2026 are aimed at frontier AI companies, the developers of the most capable models, which will face six duties including rapid incident notification and a new regulator with significant penalties. A business that uses AI tools stays under existing law: the Privacy Act, the Australian Consumer Law and workplace law. The practical change is in your supplier contracts.
Does software help decide things about your customers or staff? From 10 December 2026 your privacy policy may have to say so. Six questions give a likely answer.
Take the check →Seven months after "no new regulator", Canberra is creating one
On 10 November 2026 the Prime Minister announced that Australia will legislate mandatory AI safety and security protections, enforced by a new AI safety regulator with "significant penalties" for companies that fail to comply. In April the government had said existing watchdogs were enough. The question every owner using ChatGPT, Copilot or an AI phone agent will ask is whether this lands on them. The short answer is no, not directly. The longer answer is that it changes what you should expect from the companies you buy AI from.
6
Mandatory duties for frontier AI firms
PM's release, 10 Nov 2026
18
Recommendations in the rapid review
Into the OpenAI agent incident
30 Nov
Parliament's AI committee reports
Final report due
What the government announced
The Prime Minister's release says the rules will be legislated through Australian AI Standards legislation and will apply to frontier AI companies that want to operate in Australia, meaning the developers of the most capable models. Those companies will have to maintain security controls across development and deployment, identify and manage serious risks in their models, allow independent evaluation, notify the government of incidents rapidly, publish their safety and security frameworks, and keep up as standards change. The new regulator will work with the Australian AI Safety Institute. Detail is promised "in due course" after consultation, so there is no start date yet.
The trigger was the OpenAI agent that reached a Medicare statistics portal in June, which the government was not told about for months. The same day, the government released a rapid review of that incident. The Register reports it makes 18 recommendations, including mandatory reporting of incidents by AI companies and a possible "failure to prevent" offence for a company whose model causes harm when it did not take reasonable steps to stop it. Those are recommendations for further work, not law.
Does the AI safety regulator apply to small business?
As announced, no. A plumbing business that uses an AI quoting tool, or an accounting firm that drafts client letters in Copilot, is a user of AI, not a frontier developer. Nothing in the release puts a new duty on you, and it removes nothing either. Your AI use is still governed by the laws we set out in our guide to AI compliance in Australia: the Privacy Act, including the automated-decision disclosure duty from 10 December 2026, the Australian Consumer Law for anything an AI tells a customer, and workplace law for anything it decides about staff.
What does change is the supply chain above you. When the regulator is running, the companies whose models sit underneath your software will have to report serious incidents to the government quickly. They will not have to tell you. Most SMEs buy AI through a reseller or a software product, such as a job-management system, a practice-management suite or a booking bot, and the incident duty stops at the frontier lab. If the model under your tool misbehaves, the government may hear within days while you hear nothing, and the customer complaint still comes to you. We covered why the liability lands on the business that deployed the AI, not the vendor that built it.
If you are not sure which of your tools run on which models, start there. Map where AI already reaches into your business before deciding what to fix first.
What to do this quarter
Treat the frontier rules as a floor your suppliers should match, not a ceiling. Three steps cost little and put you ahead of whatever the final legislation says. First, list every AI tool in use and the company behind it, including tools staff signed up for themselves. Second, at the next renewal, or sooner, ask each supplier in writing which model they use, how quickly they will tell you about an incident that affects your data or customers, and what they will do to help. A supplier that cannot answer is telling you something. Third, write down who in your business owns each tool and what you would do if it stopped working tomorrow.
None of this is a reason to pause your AI plans. Mandatory rules for the companies that build the models make the tools you buy safer to rely on. Better information from your suppliers is what lets you use them with confidence.
Key takeaways
Common questions
Does Australia's new AI safety law apply to small businesses?
Not as announced. The mandatory rules and new AI safety regulator announced on 10 November 2026 apply to frontier AI companies, the developers of the most capable models. A business that uses AI tools remains governed by existing law, including the Privacy Act and the Australian Consumer Law.
When does the new AI safety regulator start?
No date has been set. The government says it will legislate the protections through Australian AI Standards legislation and release further detail after consultation. Parliament's Joint Select Committee on AI is due to deliver its final report by 30 November 2026.
Will my AI supplier have to tell me about an incident?
Not under the announced rules. Frontier AI companies will have to notify the government rapidly, but nothing requires them or the software vendors built on them to notify their business customers. Ask your suppliers to commit to a notification timeframe in writing.
Sources
Assumptions & methodology
- Last reviewed 12 November 2026 · general information for Australian businesses, not legal advice. The legislation has not been drafted; obligations may change after consultation.
- The six obligations, the new targeted AI safety regulator working with the Australian AI Safety Institute, "significant penalties" (no amounts given) and the statement that further details will come "in due course" are from the Prime Minister's media release of 10 November 2026. The release describes the rules as applying to frontier AI companies that wish to operate in Australia and does not mention businesses that deploy AI.
- The rapid review's 18 recommendations, including mandatory incident reporting by AI companies and a possible "failure to prevent" offence, are as reported by The Register on 11 November 2026. We could not access the review document itself.
- The 30 November 2026 reporting date for the Joint Select Committee on Artificial Intelligence is from the committee's terms of reference on the Parliament of Australia website. The April 2026 "no new regulator" position refers to the government's response to the Senate Select Committee on Adopting AI.
- The point that the incident duty runs to government and not to business customers is our reading of the release, which describes notification to the government only.
If this applies to you
Do you know which AI models sit under your software?
Most owners don't, and that's where the gap is. A short call maps the AI tools in your business, the companies behind them, and the contract terms worth asking for at your next renewal. Book a call.