Fighting Ai With Ai, Finance Firms Prevented $5 Million In Fraud - But At What Cost?

4 hours ago

When astir group deliberation of AI, nan first point that astir apt comes to mind isn't superintelligence aliases nan committedness of agents to boost productivity, but scams.

There've ever been fraudsters among us, that mini percent of nan organization who'll usage immoderate intends disposable to swindle others retired of their money. The proliferation of precocious and easy accessible generative AI devices successful caller years has made specified nefarious activity exponentially easier.

Also: Meet ChatGPT agent, a caller AI adjunct fresh to transportation retired analyzable tasks for you - effort it now

In 1 memorable incident from early past year, a finance worker astatine a patient based successful Hong Kong wired $25 cardinal to fraudsters aft being instructed to do truthful connected a video telephone pinch what they believed to beryllium institution executives, but were successful truth AI-generated deepfakes. And earlier this month, an chartless statement utilized AI to imitate nan sound of US Secretary of State Marco Rubio connected calls that went retired to a fistful of authorities officials, including a personnel of Congress.

And yet, counterintuitively, AI is besides being deployed by financial services companies to prevent fraud.

In a caller study conducted by Mastercard and Financial Times Longitude (a trading agency and a subsidiary of Financial Times Group), 42% of issuers and 26% of acquirers said that AI devices person helped them to prevention much than $5 cardinal from attempted fraud successful nan past 2 years.

In nan financial sector, an issuer is simply a patient that provides debit aliases in installments cards (think Chase aliases different awesome bank), while acquirers are those that judge payments (think Stripe and Square).

Also: Anthropic's Claude dives into financial analysis. Here's what's new

Many of these organizations person begun utilizing AI devices to heighten their integer information successful conjunction pinch much accepted methods, for illustration two-factor authentication and end-to-end encryption, according to a report of nan study findings published past month.

Survey respondents reported utilizing a assortment of AI-powered techniques to boost their cybersecurity and protect against fraud. The astir commonly cited method was anomaly discovery -- that is, an automated siren that flags different requests. Other use-cases included scanning for vulnerabilities successful cybersecurity systems, predictive threat modeling, "ethical hacking" (another shape of searching for strategy vulnerabilities), and worker upskilling.

The immense mostly of respondents (83%) besides said "that AI has importantly reduced nan clip needed for fraud investigation and resolution," while reducing customer churn. Even much (90%) agreed that unless their usage of AI for fraud prevention increases successful nan coming years, their "financial losses will apt increase."

Also: Researchers from OpenAI, Anthropic, Meta, and Google rumor associated AI information informing - here's why

Several barriers, however, are preventing nan financial services companies surveyed from adopting fraud-preventing AI devices astatine scale. Chief among these are nan method complexities of integrating caller AI systems pinch existing package and information that's already deployed wrong an organization. That's intimately followed by concerns astir nan accelerated gait astatine which fraud strategies themselves are evolving, which galore fearfulness will quickly outpace immoderate effort to usage AI-powered fraud prevention.

Want much stories astir AI? Sign up for Innovation, our play newsletter.