National Australia Bank Ltd. is staying with US-built AI models for now, as its technology chief says Chinese systems may lack the ethical restrictions that are placed on the largest US developers.
The US government is actively involved in making sure that Alphabet Inc.’s Google, OpenAI and Anthropic PBC do not allow their AI models to be used for cyber attacks and the firms fix such problems when they happen, said Patrick Wright, the bank’s group executive for technology and enterprise operations.
“The real concern is with everyone else, because not everyone will have the same ethical overlay in their models that the large providers are being asked to,” he said.
Sign up to The Nightly’s newsletters.
Get the first look at the digital newspaper, curated daily stories and breaking headlines delivered to your inbox.
By continuing you agree to our Terms and Privacy Policy.
If NAB looks beyond US providers, it is likely to turn to open source models before the Chinese ones, Wright said. “You never say never. You don’t know what’s gonna happen and how the world’s gonna evolve,” he said.
“Right now, we’re content to use the US-based models.”
The comments from one of Australia’s biggest lenders show the considerations being weighed by financial firms as many of them push into deploying the technology.
A flurry of model launches from China’s AI sector is seen to pose a challenge to its US rivals, showcasing the Chinese firms’ advances in high-end capabilities of reasoning, coding and complex tasks, at a lower cost to customers.
NAB is currently using agentic AI for some internal processes, including helping its operations team complete complex tasks faster.
Eventually, the Melbourne-based firm wants AI agents to interact with its staff and systems on behalf of customers, with pilots planned over the next year, according to Wright.
NAB ranked 34th in the 2025 Evident AI Index of global banks listed by AI maturity, ahead of rivals ANZ Group Holdings Ltd. and Westpac Banking Corp., but well behind Commonwealth Bank of Australia, which placed fourth.
The country’s largest lender CBA has rolled out new AI services this year and Chief Executive Officer Matt Comyn said agentic systems were helping write fraud rules and strengthen security patching.
Testing at NAB will focus on reducing the potential for unforeseen errors, given the focus on security, Wright said.
“There’s enormous potential and value for our company, but we want to make sure we do it in a safe way,” he said.
