How cyber criminals use ChatGPT to make better scams

Ben Doyle, the chief information officer at defence and transport manufacturer Thales Group, said cybercriminals are already adopting AI tools such as ChatGPT “where they can”.

“When you look at spam and phishing and things like that, the writing style is a lot better because they are all using ChatGPT to write the phishing links.

“They’re not just doing it in English anymore, they’re also automatically translating it, so a lot more countries with different are actually seeing an increase in phishing [attacks]. It used to be focussed on English, now they’re doing it in Spanish, Vietnamese, because it’s easier for them to translate, giving them a greater threat capability,” he said.

Paul Jevtovic, the chief financial crime risk and group money laundering officer at the National Australia Bank, said even when the victim of a phishing attack suspected they were under attack, cyber criminals were using AI-tools to stay a step ahead of them.

“The business email compromise is so sophisticated now that even when an individual suspects that they may have been compromised, and they go to make a phone call, organised crime can mimic in under 30 seconds via tools that exist to them now.

“Even the ability to check when you’re being compromised, organised crime is already a step ahead of us,” he said.

But AI isn’t just a tool for criminals, said Cyber Security Minister Clare O’Neil.

“Machine learning and AI are going to create more pervasive and complex threats, but they’re also going to build new tools to help us manage them,” she said.

Rapidly sharing threat information across business and government will be a core part of Australia’s 2023-2030 Australian cybersecurity strategy, and AI-based tools that can detect threats “at wire speed” would be an integral part of that response, she said.