PAYMENTS
this with additional actors, and a liability framework needs to adapt to this.”
There is, however, a clear distinction to be made between fraud and mis‐briefed agents. Agnes adds:“ If someone hijacks or scams an agent, that’ s fraud – just as it would be today – and consumers should still be protected. The imperative is to remediate as quickly and efficiently as possible.
“ If the agent had permission to act but got the brief wrong – bought the wrong size, for example, or ignored a price limit – that’ s where the companies behind the agent may need to step in. Instances of misinterpretation of an agentic request will require a more complex framework than today, where participants in a transaction are incentivised to do their part effectively whilst protecting the customer. We call this verifiable intent and we’ re committed to collaboration with banks, retailers and AI companies to get to the best solution.”
On Mastercard’ s own role, Agnes is clear:“ Our role doesn’ t change,” she says,“ we provide the trust layer, helping transactions happen securely, transparently and with consumer control. As AI agents begin shopping and paying on people’ s behalf, this means building the infrastructure that allows them to do so safely and in accordance with what consumers want. That’ s why we launched Agent Pay last year and followed it with Agent Pay for Machines in June, aimed at the highfrequency payments happening between systems ratherthan a single consumer checkout.”
fintechmagazine. com 91