Starling Bank is using AI to help customers spot the warning signs of purchase scams. The new tool, Scam Intelligence, lets customers upload images of items and ads from online marketplaces, and analyses them for signs of fraud before serving personalised guidance in seconds. It is available today for personal, joint and business account customers.

It’s the first time a UK bank has offered an AI-powered scam detection tool in its app, and follows the launch of Starling’s Spending Intelligence tool in June, which marked Starling as the first UK bank to put AI in customers’ hands.

“With Scam Intelligence, Starling customers can better protect themselves from scams, and learn more about the warning signs too,” said Harriet Rees, CIO at Starling. “Knowledge is power when it comes to managing and protecting your money, and we believe AI is giving our customers exactly that.”

The fight against fraud

Scam Intelligence launches as fraud cost the UK £1.17 billion in 2024, with authorised push  payment (APP) fraud, where customers unwittingly approve payments to scammers, accounting for £450 million. The bank aims for Scam Intelligence to reduce instances of APP fraud by warning customers against specific indicators of scams and educating them about what to monitor for in future transactions. 

Customers can upload an image to Scam Intelligence, including images of items, listings and messages from sellers on online marketplaces. The tool can also analyse screenshots of texts and messages from potential scammers that are asking Starling customers to transfer money.

If a customer is trying to buy a bike from Facebook Marketplace, Scam Intelligence might inform them that the price is too good to be true, the image in the ad isn’t genuine, or the bank account details don’t match the seller’s details. If a customer is trying to buy a guitar on eBay, Scam Intelligence might flag the seller’s refusal to use eBay’s secure payment features. Similarly, if customers receive requests to transfer money before seeing an item in person, Starling’s tool could detect pressure and urgency tactics.

Armed with this intelligence, customers can then decide whether to continue with the purchase, or not.

“Fraud is ruthless, and it's vital that every organisation supports people to avoid scams. I welcome Starling's new tool, which is a great example of how AI can be used in the battle against fraud,” said Minister for Fraud, Lord Hanson. “This government is stepping up our response, bringing forward an expanded Fraud Strategy to disrupt criminal networks and protect the public. I continue to urge everyone to follow Stop! Think Fraud advice.”

“We're on a mission to help our customers be 'Good with money', from how they spend it to how they save and protect it. While we work with social media platforms and mobile networks to combat fraud at source, Scam Intelligence is a tool people can use to help protect themselves.” said Catherine Britton, Head of Fraud Risk at Starling Bank. 

Starling’s AI vision

Scam Intelligence is engineered using Google’s Gemini models, running on the secure and scalable Google Cloud platform. This allows the tool to understand the context of the uploaded images and text, before Starling’s proprietary systems provide a risk assessment.

In line with Starling's commitment to privacy, customers have to opt in to use Scam Intelligence and use of the tool is not mandatory. All data remains securely within Starling’s Google Cloud environment and is not used for training purposes. 

“Fraud is a huge problem for UK consumers, and we believe AI can be a powerful tool to help tackle it,” said Graham Dury, Director, FSI, Google Cloud in the UK and Ireland. “By working together with Starling and using our Gemini models on a secure cloud platform, we’re giving customers a new way to spot the signs of a scam. It’s a brilliant example of using AI to help protect millions of people.”