NatWest trials home banking via Google Assistant smart speaker

Five hundred NatWest customers will be able to use voice recognition combined with PIN to confirm their identity

NatWest Bank is running a trial with 500 customers to test out the Google Assistant smart speaker for basic banking services.

The trial will use a combination of voice recognition and a six-figure PIN to authenticate users, enabling customers to find out the details of their bank balances, pending transactions, and recent spending - and nothing else at this stage of the trial.

Because recent reports have suggested that voice recognition alone could be an unreliable form of authentication, users will additionally need to authenticate by barking out two digits from a six-digit code, provided exclusively for voice assistant banking.

NatWest has produced a video demonstrating how the system works from a consumer's point of view. "We are exploring voice banking for the first time and think it could mark the beginning of a major change to how customers manage their finances in the same way mobile banking made a huge impact," Kristen Bennie, the bank's head of ‘Open Experience', told The Guardian.

While the system might sound unnecessarily convoluted for customers used to banking online or using banking apps, it could help people with disabilities. "This technology will make it easier for people to bank with us and could bring particular benefits to those who have a disability as voice banking eliminates the need for customers to use a screen or keyboard," added Bennie.

At a later date, the ability to pay bills and transfer money is likely to be added.

It's far from the first use of biometrics to banking, despite the decision by the likes of Nationwide not to forge ahead with iris recognition on cash machines due to fears of customer squeamishness. It is, nevertheless, continuing work on other forms of biometrics.

In 2016, global bank HSBC rolled-out a mobile banking system that used a combination of Apple's Touch ID fingerprint scanner and voice recognition to authenticate users.

In 2017, researchers at the University of Eastern Finland suggested that voice recognition systems could be easily fooled by impersonators. The study indicated that cyber criminals could use different technologies to fool the voice recognition software, such as voice conversion, speech synthesis and replay attacks - or just impersonating the voice of someone else.

But in less than ten years time, AI and voice-recognition powered smartphones will be more literate than a large chunk of the UK adult population, according to a report by the University of Massachusetts Amherst and Project Literacy.