Banks face agentic AI challenge in future payments

Traders Union
All News
Financial News
Banks face agentic AI challenge

Highlights

Agentic AI tools like Meta's Muse, with over 6.6 million downloads in two weeks post-September 8 launch, are rapidly changing customer-bank interactions beyond productivity gains.
Legacy anti-fraud systems struggle with AI-driven transaction patterns, risking operational disruption as agents execute thousands of authorised payments across currencies and merchants, especially for small businesses.
Banks are urged to collaborate with AI developers and regulators to safely integrate authorised agents, as UK AI minister Kanishka Narayan warns cyber attacks by rogue AI agents are 'likely, if not inevitable'.

Agentic AI reshapes banking model

As reported by the Financial Times, the debate over artificial intelligence in banking is shifting away from cost cutting and towards whether AI agents will become the main interface for customers making purchases and managing financial tasks. The argument is that banks risk focusing too narrowly on productivity gains while a broader change in customer behaviour is already taking shape.

The article says agentic AI can already research products, compare prices, organise diaries, send emails, book travel and handle multi-stage tasks for users. It points to rapid consumer uptake of such tools, citing Meta's Muse, described as a "human concierge", which secured more than 6.6 million downloads and topped the U.S. app charts within two weeks of its September 8 launch.

That transition remains incomplete because many services still require a customer to finish the payment manually. The article argues that a fully autonomous model would require AI systems to understand user preferences, operate within explicit permissions and complete purchases without forcing consumers back into banking or merchant apps.

Fraud controls and industry response

Banks and payments companies now face a conflict between longstanding anti-fraud systems and the rise of non-human transaction activity. Existing protections are largely built around verifying human intent, with layers of controls designed to block bots and suspicious automated behaviour.

The pressure could intensify at scale because AI agents do not work to human patterns and may carry out thousands of transactions across merchants and currencies, especially for small businesses. Systems calibrated to normal human conduct could therefore flag authorised agent activity as suspicious, creating operational and regulatory challenges for the sector.

The article argues that banks, whether incumbents or neobanks, need to work with AI groups such as OpenAI and Anthropic as well as regulators to define how authorised agents can operate safely. It adds that UK AI minister Kanishka Narayan this week warns that cyber attacks on the UK by rogue AI agents are "likely, if not inevitable", underlining the need for a framework that keeps out malicious actors while allowing approved AI tools to act for customers.

Our earlier article covered U.S. regulators’ $1.2 billion penalty against Bank of America after finding major failures in its anti-money laundering and sanctions controls, including billions in suspicious transactions that were not properly reported. We also noted that the bank was put under stricter oversight orders requiring remediation and third-party support, highlighting how regulators are intensifying scrutiny of banks’ control frameworks and management accountability.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.