Insights

Credentials Before Agents

What the Canada-EU summit agreed on digital trust, and why the order matters for AI agents and instant payments

Credentials Before Agents. What Canada and the EU agreed on digital trust. Wendy MacKinnon. Trusted credentials before money moves.

Most of the coverage of this week’s Canada-EU summit in Montréal was about a label. Prime Minister Carney says Canada is working to become the European Union’s first associate member. The joint statement does not use the term, and several member states, Germany most visibly, have pushed back on it. The two sides now have until next year’s summit to define what an “ever-closer association” means.

The label matters. But the practical work is further down the joint statement, in paragraph 26.

What was agreed

Paragraph 26 commits Canada and the EU to work toward mutual recognition in three digital areas:

01
Digital identity, digital credentials and trust services
02
Conformity assessment results for AI systems
03
Cybersecurity

This builds on a memorandum of understanding on digital credentials, digital identity wallets and trust services that the two sides signed through the Canada-EU Digital Partnership Council. It sits alongside the new Digital Trade Agreement, concluded in substance at the summit, which covers e-signatures, e-contracts and cross-border data flows.

The same statement commits both sides to support work toward connecting their instant payment systems. The backgrounder names the Eurosystem, the Bank of Canada and Payments Canada, and describes linking Canada’s planned Real-Time Rail to the EU’s TARGET Instant Payment Settlement system. A joint working group on banks, insurers and fintechs is to launch after the summit.

Why the order matters

Put those two commitments side by side. One connects the money. The other connects the trust.

Instant payments settle in seconds and are hard to reverse. That is the point of them. It also means the check on who is on the other side of a payment has to happen before the money moves, not after. When the payer and the payee are in different jurisdictions, that check depends on each side accepting the other’s credentials. Mutual recognition is what makes that work at scale.

Now add AI agents. A customer’s agent buying from a merchant in Lyon, or a business’s agent paying a supplier in Hamburg, does not size up a counterparty the way a person does. It needs something it can verify: a credential issued by a recognized authority, checked against a trust framework the customer’s own jurisdiction accepts. Without mutual recognition, the agent is left to trust whatever the counterparty says about itself.

That is why I read paragraph 26 as the foundation for the rest.

Credentials before agents.

What I will be watching

The statement is a commitment to work toward recognition, not recognition itself. The details will decide how useful it is:

Where Twinvocate fits

At Twinvocate, we are designing a customer-loyal agent around one rule: check who you are dealing with before you commit the customer’s money. The model proposes. The code commits. Cross-border credential recognition is the public infrastructure that rule relies on, and it is encouraging to see Canada and the EU put it in writing.

I would welcome hearing from others working on digital identity, trust services or real-time payments on either side of the Atlantic.

Sources

Wendy MacKinnon is the founder and CEO of Twinvocate, a Toronto company building a customer-loyal AI agent and a certified-counterparty credentialing layer for agentic commerce.

All insights