How AI agents could blindside the banking industry

This article summarizes an episode of Fintech Fireside Asia’s video series  featuring Zac Cohen, chief product officer at Trulioo.

Image credit: Arsal Ysfin

The next major tech wave will feature AI software that spends money and makes purchases for users. Zac Cohen, the chief product officer at identity verification platform Trulioo, watches digital fraud trends for a living.

He warns that as these digital helpers begin moving money without human oversight, the financial industry faces a huge risk. Autonomous AI, he believes, could rapidly outgrow current, outdated corporate security systems.

The false belief of having more time

This huge risk is growing because banks currently treat automated payments as a distant future experiment, assuming they have years to build safety rules. However, consumers will shatter those schedules.

Once reliable automated buying tools arrive, shopping habits will change instantly. Cohen says that while, “automated shopping is currently under 1% of transactions, it could jump to 25%, 50%, or even 75% as soon as next year.”

He warns, “The industry might not be prepared for how fast we think that change is going to happen… Studies will say that around 25% or so of consumers are comfortable with an agent autonomously acting on their behalf.”

The fight for market control

When shopping habits change overnight, the fight for market control will be ruthless. Traditional payment companies falsely assume their current users guarantee future revenue.

These companies ignore how automated software will stand between the shopper and the bank. Banks that fail to build the necessary security layer will be reduced to basic, powerless pipes.

Cohen warns, “If you’re solving for that trust layer, I think that it could take off and have a very serious first-mover advantage. The laggards are not only going to lose potential revenue opportunities, but they might be disintermediation completely.”

Choosing features over basic safety

Companies must focus heavily on security, but many software builders do the exact opposite. Software builders usually judge success by coding speed.

Product managers aggressively prioritize launching cool new features, treating safety as an annoying problem to fix later. This rush creates dangerous blind spots:

• Software builders choose fast launches over safety.
• People get angry instantly when automated security fails.
• Old rule-checking tools use slow human reviews instead of fast computer checks.

“It’s not a feature issue that we’re solving for anymore,” Cohen says. “It’s really the governance and trust framework… Payments start with features and capability, but then [the focus] very quickly goes to trust, governance, and high-volume usage.”

Finding the right owner for automated security

Moving past physical papers

The reality of checking software

The changing threat of digital crime

The gap between government rules and working code