
Open Banking Expo’s Global Regulator Day brought regulators and policymakers from around the world together for seven sessions ahead of the main show. The topics ranged from cross-border infrastructure to consumer incentives, but the liability question kept surfacing and no regulator claimed to have the answer.
The FCA's answer: not yet, and it will depend on the use case
Andrew Self, the FCA's Head of Department for Open Banking and Open Finance, was direct about where things stand on open finance liability: the answer isn't settled, attribution will likely differ by use case, and the FCA plans to publish an iterative approach around the turn of the year.
The sharpest line of the day came from the research, not the regulators
Sanya Juneja, Teresa Lam, and Bill Roberts from The Cambridge Centre for Alternative Finance presented its third major piece of work in this area, building on a 95-jurisdiction global review and a 16-country Asia-Pacific assessment, this time through 33 interviews across the countries studied. Its findings on liability are fascinating: 77% of the countries studied have more than one authority involved in open finance oversight, duties of care and consent blur as data moves from one party to the next, and liability models range all the way from single-party responsibility to fault-based attribution - the fairest model on paper and the most demanding in practice - since it depends on audit trails and signed consent actually existing when something goes wrong.
They also stated that, in practice, liability tends to land on the bank as the customer's first port of call, regardless of what the rulebook actually says. That's independent, multi-country research arriving at the exact conclusion this blog keeps describing: the party the customer knows left dealing with the situation by default, whatever the formal allocation of responsibility was supposed to be.
Even well-built cross-border infrastructure leaves the legal questions open
Hakan Eroglu, from the Bank for International Settlements, demonstrated how Project Aperta connected open finance networks across the UK, Hong Kong, the UAE, Brazil and India without changing a single domestic rule, by using a translator to convert between standards and a directory to let institutions find counterparts across borders. The technology worked, but the team called out that the legal framework, the governance model, and who actually operates and maintains the connection once it's live remain unresolved to date.
The same gap extends to AI agents
Trust infrastructure provider Raidiam's session made the point clearly: trust has to be dynamic, checked at the moment of use, not granted once and assumed to hold. That applies to people and data today, and the same model is already being extended to AI agents, each one expected to carry its own identity, purpose and chain of custody from the moment it's spawned.
What the Global Regulator Day confirmed
A UK regulator saying the liability question isn't settled yet. Global, independent research confirming liability lands on the bank by default, regardless of what the rules say. A technically successful cross-border pilot explicitly leaving the legal and governance questions for later. And trust infrastructure built for people and data already being extended to agents, before the liability question underneath any of it has an answer. None of that is solved by a better rulebook alone. It's solved by infrastructure.
Invela is the infrastructure layer that makes open finance trustworthy - accrediting who's in the network, monitoring risk in real time, and ensuring liability lands in the right place.



