Two years ago, we were using AI for research, writing, and making information easier to access. Then came more advanced capabilities — coding, building sophisticated apps and systems in minutes, automating business processes, and creating studio-grade images and videos. Now, AI systems are already making millions of payments.
Regulators meanwhile haven’t moved at a fraction of this speed. That gap could become dangerous as AI agents gain more control over money.
At United Nations Headquarters in New York, BeInCrypto recently led a regulatory panel examining that problem during the Future of Money summit. The discussion brought together Dottie Romo, Chief Risk and Control Officer at the US Internal Revenue Service; Dino Cataldo Dell’Accio of the UN Joint Staff Pension Fund; and Mastercard’s Julius Moye.
AI Payments are Already Happening at Scale
BeInCrypto’s State of AI Agent Payments 2026 research shows what machine-driven payments can already look like.
Between July 23 and August 26, researchers tracked 6.4 million x402 payment transactions carrying $119,947 across Base and Solana.
Most were tiny. 90.8% of transfers were worth less than one cent, but the AI agent recorded nearly 200 million settlement transactions since launch, according to the research.
The amounts remain small because agents are largely paying for individual digital resources, such as data, API access or computing tasks. The transaction frequency shows how differently machine commerce can behave.
Regulators May Need Algorithms Watching Algorithms
IRS Risk Chief warned that traditional oversight may struggle with that speed.
“They’re making millions of decisions in minutes,” said Dottie Romo from the IRS.
She said regulators currently rely heavily on periodic reports to find fraud, control failures and other risks. Autonomous finance could make that approach too slow.
Does that mean algorithms would need to monitor other algorithms? Romo said some form of automated supervision would likely be necessary.
“We’re not going to be able to do that in a fast enough pace,” said Romo.
She argued for more real-time monitoring across markets while keeping humans involved in important decisions.
Who Gets the Kill Switch?
Mastercard’s Moye pointed to the 2012 Knight Capital trading disaster and Terra/Luna as warnings about automated systems running without sufficient safeguards.
His model starts with machines detecting unusual behaviour and automatically containing the problem. Serious incidents would then escalate to humans.
“It’s really using AI and machines to apply the tourniquet and stop the bleeding and then have humans come in to do the surgery,” said Julius Moye, Manager at Mastercard’s Financial Crime Solutions.
Dell’Accio argued that automation cannot erase responsibility. He said accountability should ultimately trace back through three questions:
Who developed the code? Who implemented the code and who oversees the code?”
A Dystopian Reality
The uncomfortable reality is that AI is becoming a visible and aggressive part of modern finance, so much so that they are making millions in transactions every month.
Regulators cannot practically close this gap. Such speed is humanly impossible to match. So, we may soon face a new reality where regulators need machines to supervise machines.
Sounds dystopian, but it is a reality already being discussed by policymakers.
