Issue 9
Fraudsters use AI-cloned CEO voice to steal about €95 million from Intesa Sanpaolo's Fideuram
Fraudsters using an AI-cloned CEO voice and a fake WhatsApp message obtained about €95 million from Fideuram, the private-banking subsidiary of Intesa Sanpaolo; reports began appearing on 28 September. Some accounts put the loss at €88 million, and the investigation is working to identify the perpetrators.
Who is affected and how
- banking professional
- For large transfers, an executive instruction received by voice or message must be confirmed through a separate channel by calling back.
- chief financial officer
- The approval chain for exceptional transfers triggered by executive instruction should be replaced with multi-person verification that does not rely on voice.
Bank of England warns AI and debt risks together threaten financial stability
Andrew Bailey, Governor of the Bank of England, warned on 30 September that AI poses a growing threat to financial stability and that the UK must prepare for market shocks from an AI correction. The…
banking professionalCredit and capital-market exposure linked to AI should be tracked as a separate item in risk management.chief financial officerFunding and investment exposures should be stress-tested against an AI-correction scenario.deployment · globalMastercard expands Agent Pay with new trust and intelligence services
On 30 September Mastercard announced an expansion of its agent payments programme, Agent Pay, with new trust and intelligence services that give more context to agent-initiated transactions. Ant…
banking professionalAgent authentication and transaction context should be built into issuer-side decision rules.fintech companyJoining network agent frameworks may become a competitive requirement.deployment · European UnionECB to launch AI-agent payments proof-of-concept for the digital euro in January
According to reports of 29 September, the European Central Bank will examine payments by AI agents in the digital euro and launch a proof-of-concept in January. Fintech companies will also be…
fintech companyFirms with agent-payment solutions built on the digital euro should apply for the test or prepare for it.banking professionalAgent-based use cases for the digital euro may require new payment processes to be built.deployment · globalStablecoin payment layer for AI agents takes shape: Verona, Jeeves, Cloudflare
Verona has launched the first stablecoin for AI agents with $100 million in institutional launch commitments, Jeeves raised $110 million to build a stablecoin wallet and AI agents, and Cloudflare is…
fintech companyMachine-to-machine micropayments open a new market in which stablecoin infrastructure is a competitive advantage.banking professionalBanks should assess to what extent agent micropayments will run alongside traditional payment rails.Hungary
More key stories of the week
Apollo economist warns AI agents could trigger bank runs by shifting deposits
Apollo's chief economist warned on September 28 that AI assistants optimising customers' money could steer deposits towards higher-yielding fintech accounts. An Australian analysis says the roughly…
banking professionalDeposit-stability and funding-cost models should also be run against automated, yield-seeking behaviour.chief financial officerDeposit pricing and liquidity plans should include AI-driven shifts of funds as a separate scenario.fintech companyHigh-yield accounts may gain prominence through a new channel: agent-based customer acquisition.retail customerAI assistants can automatically move savings to an account with a better rate, which may benefit the customer.survey · globalIBM survey: many CFOs now lead AI strategy, but only 6% have embedded AI at scale
IBM data show that more than six in ten CFOs have taken on responsibility for corporate technology and AI strategy, but only 6% say AI is built into finance work and decisions at scale; 48% are at a…
chief financial officerAI capabilities are typically tied to specific roles and use cases, so scaling is the next step.accountantAI-driven transformation of the finance function requires broadening skill sets.deployment · United StatesRobinhood launches AI agents, weekend trading and earnings prediction markets
Robinhood announced on September 29 its own AI agents, weekend and 24/7 trading, as well as crypto derivatives and earnings prediction markets, to win over active traders. The stock rose after the…
retail customerAutomated trading tools and longer trading hours may also encourage higher-risk behaviour.fintech companyCompetition among brokerages is intensifying with AI agents and extended trading hours.supervision · United StatesFed's Waller: AI shopping changes what banks need to authenticate
Fed Governor Waller said on 30 September that AI-based shopping changes what banks must authenticate. Further analyses say banks and fintech companies must decide in real time whether an…
banking professionalFraud-detection and authentication models must distinguish between a transaction initiated by the customer and one initiated by their agent.fintech companySpending permissions granted to agents should be designed in tiers, with limits.survey · United KingdomUK launches AI scam watchlist as half of adults struggle to spot AI scams
The UK's National Trading Standards launched the AI Scam Watchlist on 30 September to fight sophisticated fraud. A survey found that half of Britons struggle to spot AI scams, and a quarter of adults…
retail customerMoney should not be transferred on request in voice or video messages from unknown sources without separate verification.banking professionalFraud-prevention communication and victim support should be tailored to the characteristics of AI scams.supervision · European UnionECB says rapid AI adoption creates systemic risk in financial markets
On 1 October the European Central Bank drew attention to AI risks for euro-area financial stability. According to the ECB president, rapid AI adoption creates systemic risk in financial markets.
banking professionalConcentrated AI use and decisions built on shared models should be recorded as risks.fraud · globalAI agents reportedly stole 600,000 credit card records at $25 per target
According to The Next Web, AI agents stole about 600,000 credit card records, with a target costing the attacker $25 on average; TNW did not independently verify the finding. Hungary's PCWPlus also…
banking professionalWith the cost of stealing card data plummeting, transaction monitoring and card reissuance policies may need tightening.fintech companyVulnerability testing of merchant and payment interfaces should be extended with agent-based attack scenarios.regulation · United KingdomUK financial regulators don't want AI-specific rules; FCA sees a leading role for UK in finance
Reports on 30 September say UK financial regulators do not want separate regulation for AI. The FCA's Colin Payne said the UK has missed the boat on AI but can lead in finance.
banking professionalExisting regulation applies to AI use, without a separate AI law.fintech companyThe favourable regulatory stance may count in favour of UK financial AI development.case · AsiaAI tools suspected in hack of South Korea's Shinhan Bank
Yonhap news agency reports that AI tools are suspected of having been used in the hack of Shinhan Bank. The news appeared on October 2.
banking professionalBank defences must account for intrusions accelerated by AI tools.oversightAustralian regulator to formally review banks' use of AI and its customer impact
Australia's corporate regulator said on 30 September that it will formally review the banking sector's use of AI in dealings with customers. The review will also cover the impact on customers.
banking professionalDocumentation and customer-impact assessments for AI tools used in customer communication should be prepared.supervision · United StatesCSBS announces AI supervisory framework for state-chartered banks and nonbank institutions
The Conference of State Bank Supervisors (CSBS), the US association of state banking regulators, announced on 28 September an AI supervisory framework for state-chartered banks and nonbank financial…
banking professionalState-chartered banks need to assess whether their AI use meets the new supervisory expectations.fintech companyBecause the framework also covers nonbank institutions, AI use will have to be documented.survey · global76% of treasury teams hit by fraud as deepfake attacks rise
The 2026 index from Treasury Dragons and nsKnox found that 76% of treasury teams have been hit by fraud, while the number of deepfake attacks is rising. The survey was published on 1 October.
chief financial officerPayment approval processes should be strengthened with multi-step verification that withstands deepfake attacks.banking professionalPayment-confirmation services for corporate clients should be updated according to deepfake risk.