AI bosses called to Senate probe + Surcharges are going. Will prices drop?

Sacha Barbour Gatt (0:08)

Hi, Sacha Barbour Gatt with you. Happy Monday and welcome to The Briefing. The days of surprise payments when tapping your card are almost over, with surcharges being axed later this week. The RBA says it will save us $1.6 billion a year, but it might not be that simple.

Joel Gibson (0:25)

It will actually take a while for consumers and businesses to get used to this change, but I think the next thing that they will be looking out for is those sneaky surcharge fees by another name.

Sacha Barbour Gatt (0:34)

That breakdown on the changes to how we pay with card is coming up in the second half of this episode. First, it's time for the headlines with Chris Spyrou on Monday, the 28th of September.

Chris Spyrou (0:47)

Morning Sacha and apologies to our listeners for my voice, I became a godfather on the weekend.

SPEAKER_4 (0:51)

Congratulations. Thank you so much.

Chris Spyrou (0:53)

We'll start with some news in tech this morning and OpenAI says it's hit pause on training its latest AI models after a string of its AI agents went rogue. A quick explainer, AI agents are the bots that go out onto the internet and do things for you. And it was on Thursday that we learnt one of OpenAI's got past the security on a Medicare statistics portal. And that happened all the way back in June. The company took almost three months to tell the Australian government. Since then, it's been revealed hundreds of OpenAI agents spent almost a week trying to get PBS and aged care data. They also went after New South Wales crime statistics and several US government sites. Authorities say no personal information appears to have been taken.

Sacha Barbour Gatt (1:35)

OpenAI says it's now warned dozens of organizations that its bots got around their security, including by using leaked passwords and posting info to third-party websites. It won't start retraining until it's confident it has better safeguards. And it says it will continue to hit pause when new issues arise. CEO Sam Altman admits the company hasn't as fast as it would have liked in telling people, as he and the head of Anthropic, Dario Amode, have been asked to front a Senate inquiry in Canberra this week to...

Sarah Hanson-Young (2:05)

Explain what indeed happened and how they're going to work with the Australian government and the Australian people to avoid this ever happening again.

Sacha Barbour Gatt (2:14)

That's Green Senator Sarah Hanson-Young, who's running the inquiry into AI and data centers on Thursday. She says Australians deserve to know what these companies are doing with our data, our water and our electricity.

Chris Spyrou (2:26)

Now neither company has confirmed they'll show up and experts don't expect they will. As to what the major parties are saying, both want AI training done here in Australia and they say that we do need to continue working with these companies.

SPEAKER_6 (2:40)

There is no putting your head in the sand and the more we engage with it, the more we have agency and respect of it.

SPEAKER_7 (2:45)

I'm not entirely sure who it is that they're going to put in cuffs and drag through the city streaks.

Chris Spyrou (2:51)

That was Deputy Prime Minister Richard Miles on Sky News there and Deputy Liberal Leader Jane Hume on ABC's Insiders as well.

It's interesting because the government is undertaking this forensic investigation into the rogue open AI agent and Monday's meeting of Federal Cabinet. So that's today's meeting of Federal Cabinet. We'll discuss the fallout. So we're going to be expecting more information to be coming out. Something I found interesting is that Australia's former chief UN cyber negotiator, what a title, she's come out to say that Australia's systems, like the government systems at the moment, are very old and archaic and it means that they are more susceptible to rogue agents like this coming in and not wreaking havoc, but gaining access. And then like we heard, they're scraping different bits of the internet and then posting it to third party sites. Like it is concerning, it is alarming. Of course.

Sacha Barbour Gatt (3:40)

And they've described this as benign, this one as benign. That doesn't mean the next one won't be serious.

This is showing what they can do. And also the fact that a company was able to keep it secret from the Australian government for three months is another concern as well.

Chris Spyrou (3:53)

Yeah. And we also heard that these agents are capable of writing code. I've seen them write code. We know that they wrote code or tried to write code in some of these Australian government sites. They just don't know what the code said and what its function was. So yeah, I mean, we're going to be getting way more developments out of this, but I think it is genuinely time that we have a real conversation about, yeah, we're investing in this tech, but do we need to be?