Wells Fargo: whatever your AI agent does, you did
From November 24, three days before Black Friday, Wells Fargo's updated online access agreement changes who carries the risk when an AI agent uses a customer's account. Under the new terms, reported by Tech Times, any instruction or transaction an agent submits is “deemed to have been initiated and authorized directly by you, without further verification.” Using an agent is at the customer's own risk, and customers who switch one off must keep evidence that they did and remain responsible until they tell the bank, as American Banker reported. In short: whatever your agent does, you did.
It's the liability transfer Jamie Dimon warned about on JPMorgan's earnings call, which we covered recently, now written into a contract and timed for the busiest shopping week of the year.
Why the bank wants it this way
The reason is technical. When an agent logs in with a customer's username and password, the session looks exactly like the customer at the keyboard. The bank can't tell the customer, an agent the customer authorized and a rogue agent using stolen credentials apart. Wells Fargo says as much: when customers switch agents off outside the bank, it “may not otherwise have visibility” into when that happened. Faced with risk it can't see, the bank has written it out of its side of the contract.
It's the flip side of a point University of Wyoming law dean Julie Hill made to American Banker: if a bank can't show a transaction was authorized, its liability goes way up. Wells Fargo's answer is to deem every agent transaction authorized, and leave the customer, the one party with no way to see what the agent did, to prove otherwise.
The law hasn't settled it either
Customers may not be able to fall back on federal protections. Regulation E caps consumers' losses from unauthorized transfers, but it excludes transfers made by someone the consumer gave their access device to. Hand an agent your password, and you arguably have. The Consumer Bankers Association called this the central question for the field in a January white paper, and no court or regulator has answered it. Wells Fargo's terms apply subject to applicable law, so the contract may not be the last word. But it sets the bank's starting position in any dispute.
Liability can only land in the right place if you know who's there
None of this is a problem agents had to create. When a third party connects through a proper interface rather than a customer's password, the bank knows which company holds the connection and what it's allowed to do, and the customer can usually cancel it from their banking app. Credential sharing throws all of that away, which is why the only party left holding the risk is the customer.
Putting liability on the customer is the easiest answer, but it isn't the right one. The right one starts with bringing agents onto identified, permissioned routes, knowing which company stands behind each one and whether that company is still in good standing, and agreeing across the open finance chain who is liable before something goes wrong. That's the infrastructure Invela is building.
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.
