AI agents can handle lower-risk work, while people must own decisions that impact a customer’s financial life, Flagright Co-Founder and Co-CEO Madhu Nadig writes in a new PYMNTS eBook, “Payments Leaders Rethink What Works in 2026.”

Complete the form to unlock this article and enjoy unlimited free access to all PYMNTS content — no additional logins required.

At the start of the year, I believed every alert artificial intelligence touched in financial crime compliance needed human approval. AI could gather the evidence, summarize the case and draft the narrative. An analyst would review the result and click to close or escalate.

Compliance runs on accountability. A named reviewer on every case made that accountability easy to show.

Customer teams moved faster than the approval step allowed. Once AI took over documentation, one payments company cut the time it spends writing case narratives by 60%. A wealth management firm with more than 1 million customers cut SAR filing time by 25%. The approval click stayed on every case, but on the thousandth low-risk alert in a queue, a click carries very little judgment.

Regulators have stopped resisting AI in compliance. FinCEN and the European Union’s new anti-money laundering authority now explicitly encourage advanced technology in AML/CFT programs, while accountability for every AI decision stays with the institution. That puts the design question in the industry’s hands.

Override rates gave us a better measure of oversight. When investigators overrule an agent more often than expected, the agent’s performance is slipping.

Source link