Community Trust ScoreVerified
Gracie Lin has a clear message. The OKX executive wants accountability baked into AI payment systems from the start — not patched in later when things go wrong.
Lin’s argument is pretty straightforward: current legal structures around the world aren’t moving fast enough to keep up with AI technology. That gap is already a problem. AI agents can get hacked. They can make wrong purchases. And right now, if either of those things happens, it’s murky who actually bears responsibility. There’s no strong legal framework to point to. No clear liability chain. Just a vacuum where the rules should be.
The Infrastructure Problem Nobody’s Solving Fast Enough
The bigger issue, per Lin, isn’t just the legal side. It’s the plumbing. The infrastructure underneath AI payment systems wasn’t built for what AI actually needs. Traditional banking rails are slow. They weren’t designed for the pace AI agents operate at, and they can’t handle sub-cent payments — which Lin says are basically essential for AI-driven financial transactions to function properly.
Sub-cent payments matter more than they probably sound. When an AI agent is executing dozens or hundreds of micro-transactions, a banking system that stumbles over tiny amounts or adds delays at each step doesn’t just create friction — it kills the whole value proposition. The speed advantage disappears. The efficiency disappears. You’re left with a technically impressive system running on infrastructure that can’t support it.
Lin’s point is that adding accountability mechanisms after the fact makes this worse. Retrofitting systems that are already live is expensive, slow, and tends to create new vulnerabilities while trying to patch old ones. Build it right from the start, she says, or you’ll spend years cleaning up the mess.
That’s not a new argument in tech circles, but it’s one that financial services has historically ignored until a crisis forces the issue. Lin seems to think the AI moment is moving too fast to wait for the crisis.
Regulators and Companies Need to Move Together
Lin doesn’t let regulators off the hook here. Her view is that regulatory bodies and companies need to be working together now — not sequentially, where tech develops and then rules catch up three years later. That lag is exactly the problem she’s describing.
Collaboration, in her framing, isn’t optional. It’s the only way to get clear standards that actually work in practice. Regulators alone can’t write rules for systems they don’t fully understand. Companies alone can’t be trusted to self-regulate when liability is undefined and the financial upside is enormous. So you need both sides at the table, probably earlier than either side finds comfortable.
What that collaboration looks like in practice — she didn’t specify. No details on specific proposals or timelines. But the direction is clear enough.
And the urgency is real. AI agents handling financial transactions isn’t a hypothetical anymore. It’s happening. The question isn’t whether these systems will face edge cases, errors, or security breaches. They will. The question is whether there’s a legal and operational framework in place when that happens, or whether everyone scrambles to figure it out in real time.
What Slow Banking Rails Actually Cost AI
It’s worth dwelling on the banking rails point because Lin keeps coming back to it. Outdated payment infrastructure doesn’t just slow things down — it probably caps what AI can do in financial markets entirely. If the rails can’t move fast enough, AI agents can’t execute the way they’re designed to. The gap between what the technology can theoretically do and what the infrastructure lets it actually do stays wide.
Sub-cent transactions are a concrete example of where that gap bites hardest. Current banking systems weren’t built for that scale of granularity. They can’t handle it cleanly, and that’s a real ceiling on AI financial applications that doesn’t get talked about enough.
Lin’s broader argument kind of ties all of this together. Legal accountability, infrastructure upgrades, sub-cent payment capability — these aren’t separate problems. They’re the same problem from different angles. AI payment systems need a foundation that’s fast enough, secure enough, and legally clear enough to function at scale. Right now, none of those three conditions are fully met.
The financial industry has a habit of treating technology adoption as a product question and leaving the infrastructure and legal questions for later. Lin seems to think that habit will be costly here. Maybe it already is.
She’s not predicting doom. But she is saying the window to get this right — to build accountability in from the ground up rather than bolt it on after the fact — is open now and won’t stay open indefinitely. The systems are being built. The decisions being made today about what goes into the foundation are the decisions that will matter when something breaks.
And something will break.
Frequently Asked Questions
What is Gracie Lin’s main argument about AI payment systems?
Lin, from OKX, says accountability mechanisms need to be built into AI payment infrastructure from the start, not added after the fact, to avoid legal gaps and operational vulnerabilities.
Why does Gracie Lin say sub-cent payments matter for AI?
Per Lin, sub-cent payments are a critical component of AI financial transactions, and current banking infrastructure isn’t equipped to handle them efficiently, which slows down AI-driven operations.
