Australians struggle to spot AI scam images, study shows

Commonwealth Bank research has found Australians correctly identify AI-generated scam images only 42% of the time, despite most people saying they feel confident they can spot them.

The bank said its survey showed nearly nine in ten Australians, or 89%, reported some level of confidence that they could recognise an AI-generated scam. The testing in the research produced a different outcome. The bank said respondents distinguished between real and AI-generated images with 42% accuracy.

Commonwealth Bank said that result sat below the outcome expected from a random guess. The bank framed the findings as a mismatch between public confidence and actual performance when faced with synthetic images used in scam content.

Age differences

The research also pointed to a small difference between older and younger Australians. Commonwealth Bank said Australians aged over 65 were 6% less accurate than younger respondents when asked to tell real images from AI-generated images.

The bank said the gap indicated deepfakes can affect people across age groups. It said the results challenged the idea that older Australians face a uniquely heightened risk from the format, at least in image-based testing.

Awareness levels

Commonwealth Bank said less than half of Australians, or 42%, described themselves as familiar with AI-enhanced scams. The bank linked the figure to the spread of deepfake content across channels used in everyday communication.

The bank said deepfakes have increased across social media platforms, websites, messaging apps and telecommunication channels. It did not provide a breakdown by channel in the research summary.

The findings land as banks, telecommunications providers and digital platforms face ongoing scrutiny over scam losses and fraud attempts that increasingly use automated tools and synthetic media. In Australia, scammers have used impersonation techniques across calls, texts, email and social media, often designed to prompt an immediate payment or the disclosure of account credentials.

Fraud comments

Commonwealth Bank said scammers have started using AI in attempts to increase the credibility of impersonation and social engineering approaches.

"The findings reveal a growing gap between confidence and reality - and that gap is exactly what scammers are looking to exploit as they increasingly turn to AI to target everyday Australians and small businesses," said James Roberts, General Manager of Group Fraud, CommBank.

Roberts said people should not feel overwhelmed by the pace of technological change.

"The good news is that the steps that keep people safe don't need to evolve at the same speed as the technology does. Deepfakes might be new, but the same tried-and-tested habits - slowing down, checking details and speaking with someone you know and trust, such as a family member, remains your best defence - even against AI-powered scams," said Roberts.

Bank warnings

Commonwealth Bank reiterated customer-facing guidance on how it communicates with account holders. The bank said it will not ask for banking information such as a Client ID, password or NetCode. It also said it will not include a link to log on directly from an email or SMS.

The bank advised customers to search for CommBank in a browser or use the CommBank app for account access. Such warnings have become common among major banks as impersonation scams continue to circulate through email and text messages that mimic brand names and customer service language.

Commonwealth Bank has expanded scam-prevention messaging in step with a broader industry push that includes stronger transaction monitoring, customer alerts and collaboration between financial institutions and telcos. The bank did not outline changes to its own controls in the research summary, focusing instead on consumer behaviour and awareness of AI-enhanced scams.

The bank said the research indicated familiar habits remain relevant even as scam content shifts from obvious fakes to synthetic images and other AI-generated material. It said Australians face deepfake content across multiple channels and that the best response includes slowing down and checking details before acting.