AI Technology in Banking

Starling Bank has introduced a tool named Scam Intelligence to aid customers in identifying fraudulent activities before transferring funds. This tool is integrated into Starling’s mobile banking application and is available to personal, joint, and business account holders. It employs Google’s Gemini AI model, hosted on Google Cloud, to analyze images and text for potential scam indicators. The process is secure, with no banking data being shared or stored within Gemini.

Functionality of Scam Intelligence

Scam Intelligence allows users to upload images and text from online marketplace listings for analysis. This includes platforms like Facebook Marketplace and eBay, where the tool can detect warning signs such as suspicious pricing or fake images. After analysis, customers receive feedback to help them evaluate the legitimacy of the listing, potentially preventing authorized push payment (APP) fraud.

Need for Scam Intelligence

Fraud is a significant issue in the UK, with losses amounting to £1.17 billion in 2024, including £450 million from APP fraud. Online marketplaces have become common targets for fraudulent activities. Scam Intelligence aims to reduce such instances by educating users on identifying scams before completing transactions.

Privacy and Security Measures

Starling Bank ensures that privacy is a priority with Scam Intelligence. Customers must opt in to use the tool, and data is managed within Starling’s secure infrastructure on Google Cloud. Images are temporarily analyzed by Google’s Gemini AI model but are not stored or shared. The tool remains optional for customers, reinforcing the bank’s commitment to responsible AI use.

Implications for AI in Banking

This initiative demonstrates Starling Bank’s approach to integrating AI into banking, focusing on fraud prevention and consumer education. By proactively identifying scams, Starling distinguishes itself from traditional banks that typically rely on post-fraud detection methods.