AI is moving from pilot to infrastructure across finance this week, with banks like Bank of Ireland embedding it in fraud detection and customer guidance while deepfake and agent-impersonation threats force new identity defences. Deal teams are wiring AI directly into proprietary data, and insurers are racing to work out who carries the risk when AI agents go wrong.
Top story: World launched proof-of-human identity tools for banks at Money20/20 to counter deepfake fraud and verify AI agents acting on customers’ behalf.
Bank of Ireland embeds AI in fraud detection and customer guidance
The Irish Times · Finance
Bank of Ireland says it is ‘accelerating’ AI adoption, with the technology now embedded in its fraud detection software and in customer financial wellbeing insights. Its generative AI tool has identified €55m in potential tax refunds for customers since an April launch, and the bank appointed a chief AI officer on 1 October. It is a concrete European example of a retail bank moving AI from experiment to core service.
World launches proof-of-human tools to stop deepfake bank fraud
The Paypers · Risk
World, the identity project co-founded by Sam Altman, unveiled World ID for banking and finance, combining an identity app, Zoom verification and credentials that let AI agents prove they act for a verified person. It targets deepfake impersonation across customer accounts and employee workflows. As AI-generated identities improve, banks and asset managers will need ways to verify both people and the agents acting for them.
Dealpath opens real estate deal records to ChatGPT, Claude and Copilot
Yahoo Finance (Dealpath announcement) · Tools
Dealpath made its Model Context Protocol connector generally available, letting any MCP-enabled AI assistant search a firm’s incoming listings, recall why past deals were passed on, and turn offering memorandums into investment committee materials. Nearly 300 institutional real estate investors use the platform. It shows investment teams treating their proprietary deal history as the asset, while keeping the freedom to swap AI models. This item is sourced from a press release because editorial coverage was not available.
Insurers write AI risk out of policies as Aon flags 300 cases
Business Model Analyst · Legal
Aon reviewed more than 300 AI-related legal cases and concluded claims could land across crime, IP, media, cyber and tech E&O policies, as reported by the Financial Times. Insurers are meanwhile adding exclusions, leaving the companies deploying AI holding exposure that was never explicitly priced. This matters for any firm using AI agents, because existing cover may not respond when something goes wrong.
https://businessmodelanalyst.com/ai-insurance-gap-silent-ai-exclusions/
Copenhagen’s Plinto raises €2.9m for AI agents policing company payments
EU-Startups · Strategy
Danish startup Plinto raised a €2.9m pre-seed round led by People Ventures and Emblem for AI agents that check whether transactions match contracts, budgets and internal policies. The agents plug into ERP, banking and messaging tools such as Slack and Teams. It reflects a wider shift towards finance control being distributed across a business and coordinated by AI agents rather than owned by one team.