Your bank is already using AI. But what’s coming next could be radically new

In June 1967, the world’s first “automated teller machine” or “ATM” was unveiled at a branch of Barclays Bank in north London in a grand ceremony.

That very first system looked a bit different to the one we know and use today. But almost six decades later, it’s hard to imagine a world where people could only withdraw cash during banking hours.

Now, in Australia and around the world, banks are placing enormous bets that a new kind of automation will transform their business model: artificial intelligence (AI).

On Monday, Bendigo Bank announced it had signed a multi-year agreement with Google to use the tech giant’s Gemini Enterprise AI platform to assist with a range of tasks, including assessing loan applications and detecting fraud.

It follows a major deal between Commonwealth Bank and OpenAI, announced in August, to “bring advanced AI to customers and employees”.

What does the future of banking hold – and who is responsible for managing the risks?

Some big changes have already happened

Banks have already been quietly deploying AI tools over many years to help with a range of tasks. If you have engaged with a chatbot recently, you have more than likely engaged with AI.

Currently, AI is helping banks and employees make decisions. It is scanning for fraud and scams, assessing credit scores, supporting trading and investment activities, and handling routine, time-consuming tasks.

That warning from your banking app about a dodgy transaction? Most likely AI. The suggestion the caller claiming to be from your bank might be a scammer? Likely AI again.

At Commonwealth Bank alone, AI tools have reportedly helped cut customer scam losses by half and slashed call centre waiting times by 40%.

The banks leading this charge aren’t just Australian. US investment bank JPMorgan, for example, has developed its own proprietary AI platform, LLM Suite, which has reportedly been rolled out across its business lines to help staff with a wide range of tasks.

What’s coming next

A recent report on AI adoption by research firm Evident Insights found that currently, about 85% of banks’ current usage of generative AI is internal, not client-facing.

But the next wave of AI adoption could be fundamentally different. Instead of just helping humans work faster, the technology could be trusted to make decisions and take action on its own.

This is called “agentic AI”. While only some banks – such as Bank of New York Mellon – have tested it, the early results are promising.

Recent research by consulting firm McKinsey profiled the case study of one major global bank, which set up ten “teams” of AI agents to handle new customer applications from start to finish.

These AI agents checked government registries, verified identities, screened for sanctions, and compiled reports. Humans only stepped in for unusual cases.

The productivity gains? According to McKinsey, while basic AI automation might make a team 15–20% faster, giving AI full control could theoretically boost output by between 200% and 2,000%.

Hard lessons

Australian banks are betting heavily on this future. But they’re also learning painful lessons about the human cost. In July, 45 Commonwealth Bank call centre workers were told they’d lost their jobs after an AI chatbot was rolled out.

Then in August, after a dispute was raised by the Finance Sector Union, the bank admitted the process could have been handled better and reversed the job cuts in question.

Despite the bank’s backtrack, Commonwealth Bank Chief Executive Matt Comyn later told a technology festival in October that making the most of AI “needs to feel urgent”. He said leaders needed to take initiative, despite a temptation to sit back and follow.

What does all this mean for the future of banking?

The financial services industry is continuing to experiment with the best ways to use AI.

One option is to create AI-powered financial coaches that proactively message customers with personalised savings tips.

Another being explored includes “autonomous finance” systems that could manage your money with minimal input, optimising everything from bill payments to investment allocations.

This means that, in the near future, AI systems could run entire banking processes on their own. Imagine applying for a loan at 2am and getting approved five minutes later, with AI handling every single step.

What about the risks?

The public expects banks to deploy fair, explainable and secure AI systems. But the technology is moving so fast that regulators are scrambling to keep up.

There’s particular concern about algorithmic bias. If AI learns from historical data reflecting past discrimination, it could perpetuate or even amplify unfair lending practices.

For example, this could negatively affect borrowing ability for those historically seen as a “bad investment”.

The banks themselves are responsible for any mistakes made by AI. Accountability cannot be outsourced to algorithms. However, it is likely customers who will still feel the brunt of those mistakes.

Banking is set to be fundamentally rewritten by AI, whether we’re ready or not. That could mean cheaper, faster, more personalised banking.

But it also threatens jobs, raises privacy concerns and concentrates enormous power in algorithms most of us don’t understand.

As politicians turn up the heat on banks, the real test isn’t whether AI can transform banking. It’s whether that transformation will be fair and not just for the bottom line.