LAURA JAYES, HOST: That was Anthony Albanese live at the UN with Penny Wong looking on there. So is the Innovation Minister – Tim Ayres is in the studio with me. Good to see you, Tim.

SENATOR TIM AYRES, MINISTER FOR INDUSTRY AND INNOVATION AND MINISTER FOR SCIENCE: G’day.

HOST: He talked about the breach. This is really good timing, if not great politics, to talk about the breach that we've experienced here when you're still calling for greater AI regulation. As breaches go, this is not a serious breach, is it?

AYRES: Well, the first thing I might say is, that was a very good outline of Australia's approach, and it's a credit to your network to play the Prime Minister's address there uninterrupted. Because it does, for your viewers, give an outline of, what is the Albanese Government's approach in global politics. Labor and Australia did set up the United Nations. It is partly an Australian creation. And making sure that we are represented effectively and we've got those international relationships right.

HOST: Well, if you own the set-up, do you own what the UN has become – which is a little bit ineffective?

AYRES: Well, you would have seen that in the Prime Minister's outline. The pressure that the rules-based order is under, means more Australian agency in areas like artificial intelligence. As I neatly segue back to your original question. 

HOST: Thank you!

AYRES: That does underline why we are adopting the approach that we've adopted. More Australian agency. Onshore investment here, so that Australia is shaping the artificial intelligence technology world. Building capability within Government, to make sure that we're dealing effectively with those safety challenges. And the breach that has been outlined recently is a good example of that. 

HOST: As breaches go, it's not a serious breach. ASD didn't pick it up.

AYRES: Well, it's a serious breach that the Government is taking seriously. You're right to say that there aren't, for example, personal details of Australians’ Medicare accounts or other personal data, or secret government data that has been released here. That is true. But it's important to learn the lessons of this breach, to make sure that OpenAI learn the lessons of this too. It is unacceptable to have this kind of activity going on and where we are, as the PM outlined in New York, building a process to make sure, that within government at a technological level and a government-process level, we get this right for the future.

HOST: These breaches happen all the time, they're just not from autonomous agents. So, I put it to you, that this breach was so unserious that it was way below the threshold of what the ASD would have picked up.

AYRES: No, it's serious because it underscores this issue.

HOST: The autonomy of it?

AYRES: About artificial intelligence agents doing this kind of work.

HOST: So the autonomy, not the actual breach?

AYRES: And what that means for our artificial intelligence standards that we'll legislate. It is true that there are ever-present cyber security risks right across the Australian Government's network, and for big companies.

HOST: And ASD didn't pick this up?

AYRES: Well, ASD was notified in the timeline that you will have seen some three weeks after the breach happened. And it is, I think the way that you're approaching the question is right, the actual outcomes, the negative outcomes of this – very limited. But the event itself is unacceptable, and it is an opportunity for the AI Safety Institute, ASD, Services Australia all to be working together.

HOST: You set up the AI Safety Institute. What's it doing?

AYRES: Well, the cybersecurity approach across government is a responsibility of the individual departments, ASD, broader agencies.

HOST: The Safety Institute is not proactive, it's reactive.

AYRES: No, the AI Safety Institute's job here is to engage with the tech platforms to do testing of their models, to identify risks and to provide advice to government. They have done that in this instance. They've done it prior to here. This is an opportunity for all these parts of government to work together and –

HOST: But Tim, the AI Safety Institute does not test the models before they're in the market here. So, again, it is not proactive, it's reactive. Isn't that the number one thing –

AYRES: Again, I couldn’t ask for a more helpful segue into, what is the Government's approach here? It is to secure onshore investment in AI training in particular, because that is an important sovereign capability for Australia. And that means questions like, pre-launch testing, are absolutely in the zone of the issues that we want Government to be able to do. At the moment, all of this technology –

HOST: So, you're not really calling for regulation, you're calling for more access from these tech giants and using this massive platform at the UN to do so?

AYRES: Well, we'll be legislating AI Standards for Australia. That is a set of regulations there that don't just go to the energy questions, they do. But also making sure that we've got the capacity to keep Australians safe and build Australian agency and sovereignty and resilience in artificial intelligence. Which is a transformative technology for the world, not just in economic terms, but in strategic and safety terms as well.

HOST: Indeed, it is. Truncated today – we'll see you soon, Tim. Appreciate your time.

AYRES: Thank you.