Complete the form to unlock this article and enjoy unlimited free access to all PYMNTS content — no additional logins required.

Payments infrastructure has become an expert at answering questions centered on human behavior. Is this really the cardholder? Is this account authorized? Is the transaction fraudulent?

Artificial intelligence agents introduce another question. Even if the machine is legitimate, was it allowed to make this particular decision?

“The technology is moving faster than the liability frameworks,” Paymentology Chief Technology Officer Tim Joslyn told PYMNTS for the October edition of the “What’s Next in Payments” series, “The Last Thing a Human Will Need to Approve.”

“Our problem is not automation,” Joslyn said. “Our problem is interpreting and consuming everything it does for us in a safe way.”

That distinction also changes how enterprises should think about deploying agents. AI autonomy is frequently discussed as a capability, whereby once a model becomes sufficiently accurate, organizations can remove the person supervising it.

“For me, autonomy is earned with evidence,” he said. “What you’re not going to do is be caught up with the enthusiasm of, just because something can make a decision, it should make that decision.”

The unlock innovation across agentic payments, it turns out, looks less about building agents capable of doing everything than defining exactly what each agent is allowed to do.

” Joslyn said. “Is it the person who instructed the command in the first place?

Source link