2 min readPayment systems and banks in Australia are working to implement scam detection using AI, but our data concerns may hold them back. At a panel discussion around AI and cyber security at SXSW Sydney, the conversation kept coming back to collaboration. Between organisations, between the government and businesses, and between payment systems and their users. While the panellists agreed that Australian stakeholders did work together, there still wasn’t enough information being exchanged.
One panellist, Financial Services and Insurance Lead for CyberX, Shameela Gonzalez, said public sentiment about data privacy would need to shift to allow AI-powered scam and fraud detection to thrive.
“I would argue that public sentiment [shifting] towards “is my data being used to protect me” would actually heighten how much we would allow industry to utilise, if we fundamentally knew it was being used for protection.”
Privacy, and in particular data privacy, is at the forefront of Australians’ minds with almost half of the population being compromised in a data breach between 2022 and 2023. With data breach concerns rising by 13 percentage points to 74% in 2024, awareness and fear is on the rise, which has in part led to stronger privacy obligations for organisations. While greater privacy controls are beneficial for users, it does present a problem for those looking to train AI programs.
In the payments sector, organisations are working to create AI technologies that can detect potential scams through a system called Scam Scores. These scores look at transactions and user behaviour patterns to determine whether a payment looks suspicious. These could then prompt users to reconsider the transaction, and alert the financial institution to a potential threat. Like all AI technologies, these systems require data to train on and learn from, but without enough of it, the technology plateaus.
According to Gonzalez, “the effectiveness and the efficiency of artificial intelligence will only be as good as the data it can use… I think it's going to be a really tricky line for the industry to dance on.”
To the contrary, fellow panellist, Business Lead at Featurespace APAC Phillip Finnegan says his organisation looks to lean further into behavioural changes rather than data specifics.
“Sometimes we don’t need as much data as we think, actually. That's one of the things that often comes into debate. We think about, what is the minimum data we can actually analyse from a transaction.”
As it stands, financial institutions can take the data they need to create these AI systems, provided it is within the terms of service. A survey from OAIC suggests that while many Australians have agreed to a certain level of data collection, three in five still don’t know what organisations do with their data, and the vast majority want more control over it, not less. 90% of Aussies want to be able to object to data collection while still using a service, and 89% want to know when their data is being used in decision making that affects them.
Clearly, there is a trust gap.
Both panellists agree that transparency is key to better outcomes for businesses and consumers alike, and that the reforms to the Privacy Act will do their bit to bridge this.
Among other things, the amendments introduced to parliament in September will see better transparency on the use of automated decision making for users. They will also compel private and public organisations, and international partners to collaborate more openly where privacy is concerned.
These steps forward are small, but positive. Seemingly they check the boxes Gonzalez and Finnegan believe could have an impact on consumer sentiment, but how that plays out in the court of public opinion is yet to be seen.
