The rails. Why European agentic commerce is co-defined by two converging regimes.

📊 Full opportunity report: The rails. Why European agentic commerce is co-defined by two converging regimes. on ThorstenMeyerAI.com — validation score, market gap, and execution plan.

TL;DR

European agentic commerce is being shaped by two regulatory regimes—PSD3/PSR and the AI Act—that are developing simultaneously. This statutory approach differs from the US’s private infrastructure, affecting speed and openness.

European law is currently shaping the infrastructure for agentic commerce through two major regulatory regimes: PSD3/PSR and the AI Act, which are being developed simultaneously and will jointly define how AI agents can operate in payments and high-risk AI functions.

The core issue is that, unlike in the US where private firms build and extend payment rails, Europe’s payment infrastructure is governed by statutory regulations that require human authorization for transactions. PSD3 and the Payment Services Regulation (PSR), agreed in November 2025 and expected to be implemented by 2028, will rebuild payment rails with mandatory API parity, forcing banks to expose interfaces as capable as their own apps. Simultaneously, the EU AI Act, with high-risk obligations set for 2026, classifies AI systems used in finance—such as credit scoring and fraud detection—as high-risk, subjecting them to conformity assessments, human oversight, and registration. These two regimes are being developed independently but will jointly influence the capabilities and limitations of agentic commerce in Europe.

Thorsten Meyer, a researcher tracking these developments, emphasizes that the European approach is not merely an extension of existing technology but a redefinition of the underlying legal architecture. The convergence of these regimes means that whether an AI agent can perform payments depends on the evolving legal framework, not just technological capability. The process is slower than in the US, where private networks like Mastercard’s Agent Pay and Visa’s Intelligent Commerce operate on commercial rails that can be extended by decision. In contrast, Europe’s statutory rails are built into law, making them more durable but also slower to implement.

The Rails — Thorsten Meyer AI
RAILS
● DISPATCH / JUNE 2026
THORSTEN MEYER AI · AGENTIC COMMERCE · § 04
AGENTIC COMMERCE · 04
EUROPE / RAILS
Essay · European-Infrastructure Forensic · 2026-06-04

The rails.
Why European agentic
commerce is co-defined by
two converging regimes.

An agent that can shop cannot pay. The gap at the center of European agentic commerce isn’t a technology gap — it’s a legal one.
The AI can compare, choose, and fill the cart — but at payment, European law requires a human, not a machine, to authorize, and there’s no mechanism to treat an agent as a legal payer. In the US, agentic payments run on commercial rails (Mastercard Agent Pay, Visa Intelligent Commerce, Plaid) a few firms own and extend by decision. In Europe the rails are statutory — defined by regulation, and being rebuilt right now: PSD3/PSR (agreed Nov 2025, publishing summer 2026) with mandatory API parity, and the AI Act classifying credit scoring as high-risk. The structural argument: European agentic commerce isn’t a product shipped onto existing rails — it’s a system co-defined by two converging regulatory regimes, so the constraint isn’t the agent’s capability but the legal architecture it must run on, and that architecture is statutory, fragmented, and different in kind from the US commercial one.
can’t pay
An agent can shop but can’t pay ·
SCA needs a human payer
API parity
PSD3 forces banks to expose
first-class third-party interfaces
Aug 2 ’26
AI Act high-risk deadline ·
(Omnibus may slip it to 2027)
~2028
PSD3 full applicability ·
the clock agentic commerce runs on
THE RAILS· AN AGENT THAT CAN SHOP CANNOT PAY· THE CONSTRAINT IS LEGAL, NOT TECHNOLOGICAL· SCA REQUIRES A HUMAN PAYER · NO MECHANISM FOR AGENTS· US COMMERCIAL RAILS · EXTENDED BY DECISION · FAST, CONCENTRATED· EU STATUTORY RAILS · DEFINED BY LAW · SLOW, OPEN· PSD3/PSR AGREED NOV 27 2025 · PUBLISHING SUMMER 2026· MANDATORY API PARITY · NO MORE DEGRADED INTERFACES· DIRECT PAYMENT-SYSTEM ACCESS FOR NONBANKS · NO SPONSOR-BANK VETO· AI ACT · CREDIT SCORING IS HIGH-RISK· FOUR INSTRUMENTS · PSR / FIDA / PSD3 / AI ACT · ONE AGENT· THE FRICTION IS INTER-REGIME, NOT INTRA-REGIME· THE MANDATE BRIDGE · AUTHORIZE ONCE, DELEGATE BOUNDED ACTION· WHICH FOUNDATION AN AGENT ECONOMY PREFERS IS THE OPEN QUESTION· THE RAILS· AN AGENT THAT CAN SHOP CANNOT PAY· THE CONSTRAINT IS LEGAL, NOT TECHNOLOGICAL· SCA REQUIRES A HUMAN PAYER · NO MECHANISM FOR AGENTS· US COMMERCIAL RAILS · EXTENDED BY DECISION · FAST, CONCENTRATED· EU STATUTORY RAILS · DEFINED BY LAW · SLOW, OPEN· PSD3/PSR AGREED NOV 27 2025 · PUBLISHING SUMMER 2026· MANDATORY API PARITY · NO MORE DEGRADED INTERFACES· DIRECT PAYMENT-SYSTEM ACCESS FOR NONBANKS · NO SPONSOR-BANK VETO· AI ACT · CREDIT SCORING IS HIGH-RISK· FOUR INSTRUMENTS · PSR / FIDA / PSD3 / AI ACT · ONE AGENT· THE FRICTION IS INTER-REGIME, NOT INTRA-REGIME· THE MANDATE BRIDGE · AUTHORIZE ONCE, DELEGATE BOUNDED ACTION· WHICH FOUNDATION AN AGENT ECONOMY PREFERS IS THE OPEN QUESTION·
FIG. 01 — THE GAP · AN AGENT THAT SHOPS CANNOT PAY
The defining constraint on European agentic commerce is legal, not technical
The capability is present; the authority is absent
shop ✓
Compare, evaluate, fill the cart,
choose the best deal — capability is here
SCA
human
authentication
required
pay ✗
No mechanism to treat an agent
as the equivalent of a human payer
Strong Customer Authentication requires two of three factors — something the payer is (biometric), knows (password), possesses (a device). Each presumes a human; an autonomous agent has none in the SCA sense. Europe’s agentic-commerce bottleneck is its own payment law — a constraint that cannot be engineered around, only legislated through. The barrier is not a missing feature; it is the regime itself.
FIG. 02 — STATUTORY VS COMMERCIAL RAILS · WHY THE US PLAYBOOK DOESN’T PORT
Two foundations, different in kind
The US playbook assumes the rail’s owner sets the rule; in Europe the legislature does
US · commercial rails
Owned by networks, extended by decision
  • Mastercard Agent Pay, Visa Intelligent Commerce, Plaid
  • The rail’s owner sets the rule — extend to agents by product decision
  • Fast — moves at product speed
  • Concentrated — a few firms control access
EU · statutory rails
Defined by regulation, no owner
  • PSD2/PSD3, PSR, SCA, FIDA
  • The legislature sets the rule — no network can grant payer status
  • Slow — moves at legislative speed
  • Open — mandatory API parity, public data substrate
A US firm cannot bring Agent Pay to Europe and switch agents on — it must wait for the European regime to define how an agent authenticates, accesses data, and pays. The playbook’s central move (extend the rail by decision) is unavailable, because the rule is set by regulation. The same property that makes the EU stack slow — statutory rails — is the property that makes it open: no agent economy built on Visa’s permission is as open as one built on mandatory API parity.
FIG. 03 — THE PSD3/PSR REBUILD · THE NEW PAYMENT RAILS
The most consequential payments reform since PSD2 introduced open banking
The clock European agentic commerce runs on
Nov 27 2025
Parliament + Council reach provisional political agreement on PSD3 and the PSR
Summer 2026
Final texts expected in the Official Journal
+20 days
PSR (directly applicable) takes effect — mandatory API parity, nonbank payment-system access
~2028
PSD3 fully applicable after ~18-month transposition · the SCA rewrite lives in the PSR
Mandatory API parity means an agent gets a first-class bank interface by law — the difference between an agent that works and one quietly throttled by the bank whose customer it acts for. Direct payment-system access ends the sponsor-bank veto over fintech models. But the SCA accommodation that would let an agent pay is not yet written — it must live in the PSR, within a framework built to fight a $400B fraud problem.
FIG. 04 — THE AI ACT GUARDRAILS · THE MODEL REGIME
Running on the rails is necessary but not sufficient
The rails govern whether the agent can pay; the guardrails govern whether it can decide
The classification
Credit scoring = high-risk
Annex III loads it with conformity assessment, human oversight, registration, post-market monitoring. The heaviest tier.
The deadline
Aug 2 2026 — maybe
The May 2026 “Omnibus” proposes slipping high-risk to 2027 — not yet adopted; treat Aug 2026 as operative.
The reach
Extraterritorial
A US lab’s agent scoring a European user is in scope even if hosted offshore. The Brussels Effect, applied to agents.
The AI Act’s human-oversight requirement intersects directly with the payment regime’s human-authentication requirement: both regimes, from different directions, insist a human stay in the loop — the AI Act for the decision, the PSR for the payment. Non-compliance reaches up to 7% of global revenue. The guardrail shapes what an agent can do beyond paying — and because it reaches any system serving EU users, it shapes agentic finance globally.
FIG. 05 — THE MANDATE BRIDGE · HOW THE GAP GETS CROSSED
Not as an autonomous payer — as a bounded delegate of a human who authorized it once
The design that threads both regimes’ insistence on a human in the loop
The human · up front
Authorizes the mandate
Sets spending limits, allowed merchants, use cases — and authenticates once (satisfies SCA).
delegated,
within
limits
The agent · within bounds
Transacts inside the mandate
Acts without re-authenticating each payment — the boundaries satisfy AI Act oversight.
The mandate satisfies the payment regime’s human-authentication requirement (the human authorizes the mandate) and the AI Act’s human-oversight requirement (the human sets and can revoke the boundaries) simultaneously. For it to scale, the regimes must formalize it — the PSR’s SCA rewrite is where the legal basis would live, the AI Act’s oversight rules are where the boundary requirements would. This is the permission-and-boundary model the European approach favors over autonomous action.
Europe is betting that durable, open, publicly-owned rails produce a better agentic-commerce market than fast, concentrated, privately-owned ones — even at the cost of arriving later. Which foundation an agent economy actually prefers is the genuine open question.
Thorsten Meyer · The Rails · Agentic Commerce 04

Implications of Dual Regulatory Frameworks for European AI Commerce

This dual regulation approach means that European agentic commerce will develop within a highly structured legal environment, potentially leading to slower deployment but greater durability and openness. The mandatory API parity and open finance provisions prevent private control of interfaces and data, fostering a more competitive and transparent ecosystem. However, the slower pace may delay the deployment of fully autonomous payment agents compared to the US, where private firms can rapidly extend their networks. Ultimately, the architecture of these regulations will influence which model—speed and concentration or openness and resilience—prevails in the future of agentic commerce.

Amazon

European payment API integration tools

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Regulatory Foundations Shaping European Agentic Payment Infrastructure

The US has relied on private infrastructure, such as Mastercard’s Agent Pay and Visa’s Intelligent Commerce, which are owned and extended by decision of private firms. Europe’s approach is different: the payment rails are being rebuilt through legislation—PSD3 and the PSR—that mandate API parity and direct access for nonbanks, aiming for a more open and interoperable system. Simultaneously, the EU AI Act, agreed in November 2025 and set to take effect around 2026-2028, classifies AI systems used in finance as high-risk, requiring compliance, oversight, and registration. These developments are unfolding in parallel but are not coordinated, creating a complex, layered regulatory environment for agentic commerce.

“The European approach is not merely an extension of existing technology but a redefinition of the underlying legal architecture.”

— Thorsten Meyer

Machine Learning for High-Risk Applications: Approaches to Responsible AI

Machine Learning for High-Risk Applications: Approaches to Responsible AI

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Uncertainties in Implementation Timelines and Regime Interactions

It remains unclear how quickly the PSD3/PSR regulations will be fully implemented and whether the AI Act’s high-risk obligations will meet their projected deadlines. The interaction between these regimes, especially how they will practically constrain or enable agentic payments and AI functions, is still being defined. Additionally, the impact of potential delays or legislative amendments is uncertain, which could influence the pace and scope of European agentic commerce development.

Amazon

payment regulation compliance tools Europe

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Next Steps in Regulatory Implementation and Market Adaptation

Regulators are expected to finalize and implement PSD3/PSR regulations by 2028, with ongoing trilogue discussions on the AI Act potentially extending into 2027. Industry stakeholders are preparing for these changes by developing compliance strategies and testing AI capabilities within the evolving legal framework. Monitoring how these regulations are enacted and enforced will be crucial for understanding the future landscape of agentic commerce in Europe.

AI-Powered Fraud Detection: Machine Learning Systems That Save Millions Annually

AI-Powered Fraud Detection: Machine Learning Systems That Save Millions Annually

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

How does Europe’s regulatory approach differ from the US in developing agentic commerce?

Europe relies on statutory regulations like PSD3/PSR and the AI Act, which build a legal infrastructure that is slower but more open and durable. The US depends on private, commercial rails owned and extended by firms like Mastercard and Visa, enabling faster deployment but less transparency and openness.

When will the new European payment and AI regulations be fully in effect?

PSD3 and PSR are expected to be implemented around 2028, while the AI Act’s high-risk obligations are likely to take effect between 2026 and 2027, depending on legislative progress.

What are the main challenges for AI agents operating within Europe’s regulatory framework?

AI agents must navigate complex, layered regulations that require human oversight, conformity assessments, and registration. The legal requirement for human authorization at the point of payment also limits autonomous payment capabilities compared to private infrastructure models.

Will Europe’s approach make its agentic commerce market more durable?

Yes, because regulations embedded into law create a stable, open infrastructure less susceptible to private control, though this comes at the cost of slower deployment and adaptation.

How might these regulations influence global competition in agentic commerce?

Europe’s deliberate, regulation-driven model may set a durable standard for openness and interoperability, potentially influencing global norms, but its slower pace could disadvantage it in rapid innovation compared to the US’s faster, private-led approach.

Source: ThorstenMeyerAI.com

You May Also Like

Why AI Search Answers Need Source Discipline

Discover why strict source discipline is vital for AI search answers. Learn how transparency, accuracy, and trust hinge on citing credible sources.

When a Content Network Starts Publishing to Itself

A major content network has started publishing content across its own properties, shifting from external distribution to internal ecosystem building. This impacts control, engagement, and revenue.

The Co-Founder’s Black Hole — A Structural Read on Jack Clark’s Automated AI R&D Essay

Anthropic co-founder Jack Clark predicts over 60% chance of fully automated AI research by 2028, raising concerns about institutional capacity and future risks.

The Free-Download Question: When Running Your Own Model Actually Beats Paying

Analysis of how owning and running open-weight AI models can be more cost-effective than paying for API access, with recent technological advances supporting this shift.