When does a provider of AI agents become a regulated payment service provider? In episode #146 of “Alles Legal – Fintech-Recht kompakt”, Frank Müller explains the regulatory boundaries businesses need to be aware of when it comes to Agentic Payments. Tune in now!
Agentic Payments from the provider’s perspective
An AI agent searches for a suitable flight, makes the booking and pays for it directly. For users, this can feel like a single, seamless process. From a legal perspective, however, a number of different services and roles may sit behind it – and for the provider of the agent, the decisive factor may be how deeply its own technology is involved in the payment transaction.
In this episode of “Alles Legal – Fintech-Recht kompakt”, host Dana Wondra from Payment & Banking talks to Frank Müller about the regulatory position of the company behind the AI agent.
After the previous episodes covered topics including contract formation, authorisation and authentication, this episode focuses on a different question: when does the provider of an AI agent remain a technical service provider – and when does it provide a regulated payment service itself?
What matters is what the AI agent actually does
For regulatory classification, it does not matter whether a company describes its product as an AI agent, software solution or technical platform. What matters is the functions it actually performs.
Does the agent merely transmit data? Does it access the user’s account and initiate a payment? Or does the provider even receive funds before passing them on to merchants?
Simply initiating a payment can already have regulatory consequences – even if the provider never holds customer funds itself. If, for example, the agent accesses a person’s bank account and initiates a credit transfer, the activity may qualify as a regulated payment initiation service requiring authorisation.
If, on the other hand, the provider receives funds itself and subsequently passes them on to merchants, money remittance may become particularly relevant. Depending on the specific architecture of the business model, other payment services may also come into consideration, potentially including the issuing of a payment instrument.
When does the exemption for technical service providers apply?
Not every form of technical involvement in a payment transaction automatically means that the provider itself becomes a payment service provider. A company that exclusively provides technical infrastructure may, in principle, benefit from the exemption for technical service providers.
This may include, for example, data processing and storage, authentication or the provision of technical interfaces.
However, the actual design of the product remains decisive. Describing the company accordingly or including a clause in its terms and conditions is not sufficient if the service provided goes beyond a purely technical service in functional terms.
What role do contractual relationships play?
In addition to the technical architecture, the contractual relationships are also crucial. On whose behalf is the agent actually acting: the customer, the merchant or possibly a bank?
In certain circumstances, the so-called commercial agent exemption may also be relevant. Its requirements are narrow, however. Among other things, it must be clear which party the provider is acting for and what actual authority it has to represent that party.
Particularly with Agentic Payment models, it therefore makes sense to consider the technical processes and contractual structure together.
Stablecoins add another layer of regulation
The situation can become even more complex if the agent pays not by card or credit transfer, but using stablecoins.
The use of stablecoins does not mean that the regulatory requirements governing the payment process disappear. Instead, additional requirements arising from crypto-asset regulation may apply alongside payment services regulation.
Providers therefore need to consider at an early stage which regulatory role they intend to assume with their product.
Technical service provider or regulated payment service provider?
A key regulatory decision is therefore already embedded in the architecture of an Agentic Payment product.
Does the company remain limited to providing technical support? Or does the agent perform functions that may be attributed to the provider itself as a payment service?
The answer can determine whether a business model can be implemented without the provider obtaining its own authorisation, or whether a licence or cooperation with a licensed partner is required.
Frank Müller and Dana Wondra also discuss how the provider’s regulatory role may affect liability and where existing payment services law may reach its limits when dealing with increasingly autonomous AI agents.
In the next episode, we change perspective once again: what happens when an AI agent no longer acts on behalf of a customer, but on behalf of a bank? The focus will then be on control and responsibility, the involvement of external AI providers and the question of when a digital copilot actually becomes Agentic Banking – a boundary that will also be discussed at the Banking Exchange in November.
About this podcast
Alles Legal – Fintech Recht Kompakt delivers sharp, weekly insights into legal and compliance matters in the world of banking. (in German only)
This podcast is a collaboration between Payment & Banking and PayTechLaw.
Each Wednesday, we unpack the legal developments shaping the financial world – clearly, concisely, and without the legal jargon.
Since 2021, PayTechLaw authors and Annerton attorneys have brought depth and clarity to complex topics.
Whether it’s PSD3, DORA, or FiDA – we provide the legal context you need.
In 20 minutes. No detours.
