2 weeks ago
Mehaniq/Shutterstock
HM Revenue & Customs (HMRC), nan UK's taxation authority, has announced a 10-year woody with British tech patient Quantexa to usage AI for detecting fraud and tax return errors. The BBC has reported that nan multi-year statement will costs nan UK authorities £175 cardinal ($234 million). Under nan partnership, Quantexa will harvester information that nan HMRC has collected pinch different sources to thief it spot incidents of fraud, arsenic good arsenic to thief it hole unintentional errors. It will besides assistance HMRC pinch its customer work needs and place companies aliases individuals trying to hide fraudulent activities. In addition, nan HMRC will usage its exertion to way down morganatic payments made by taxpayers nether nan incorrect reference number.
Quantexa, founded successful London backmost successful 2016, develops AI apps and devices for information analytics and decision-making. Aside from nan UK taxation authority, it's besides moving pinch nan Switzerland-based Zurich Insurance Group Ltd. to heighten nan company's fraud detection. While nan HMRC will usage its exertion to observe irregularities successful taxation filings and payments, nan institution told nan BBC that its findings will still beryllium checked by group and that its exertion was made to "support quality decision-making, not switch it." Hopefully, that intends taxpayers won't person to look mendacious accusations of fraud based connected nan findings by nan company's AI.
"In authorities environments, AI cannot run arsenic a achromatic box," Quantexa main executive Vishal Marria told nan BBC. "Decisions request to beryllium transparent, auditable, and explainable, peculiarly successful areas affecting citizens directly." Marria besides said that Quantexa will "never return HMRC information distant from nan HMRC environment" to guarantee nan information would stay secure.
Back successful 2024, nan US Treasury Department (home of nan Internal Revenue Service) admitted that it had been utilizing AI for fraud detection. It said backmost past that it prevented fraud and recovered payments worthy complete $4 cardinal from October 2023 to September 2024 utilizing AI tech. It has since go much communal for authorities agencies astir nan world to usage AI — nan US, for instance, has signed aggregate agreements pinch companies for illustration Google, xAI, Anthropic and Microsoft to usage thei technologies.
