FIS compressing anti-money-laundering investigations from hours to minutes using AI agents is genuinely impressive. But the Forbes story misses the hard part: that speed is only valuable if your governance framework can credibly explain and defend every decision the AI made. For UK regulated firms — law practices, wealth managers, insurers, accountancy businesses — the AML case is instructive. The FCA's Consumer Duty PS22/9 and the PRA's SS1/23 framework for operational resilience demand not just that you use AI, but that you demonstrate control, transparency and auditability. Speed without governance is compliance theatre. Firms chasing AI deployment velocity without proper controls are building reputational and regulatory risk, not efficiency.

This story is part of a much larger pattern: the fintech industry has moved past asking 'should we use AI?' and started asking 'how do we make AI trustworthy enough that regulators and clients accept it?' That shift is real and necessary. What's revealing is that the answer isn't 'more advanced models' or 'better algorithms'. It's governance architecture. FIS's partnership with Anthropic works because Anthropic's constitutional AI approach and Anthropic's own transparency reporting give FIS something to point to. Meanwhile, some firms are still deploying generic LLMs like Microsoft Copilot or Harvey into sensitive workflows with minimal guardrails, hoping legal indemnity clauses will save them. They won't. The EU AI Act is coming. The ICO's AI governance guidance is crystallising. The regulatory bar is rising fast, and 'we bought an AI tool' is no longer a defence.

Here's what we think: AI agents in AML, KYC, sanctions screening and fraud detection are genuinely valuable. But they have to be implemented governance-first, not speed-first. That means: documented decision logic that auditors can inspect; explainability built into the pipeline, not bolted on afterwards; clear human sign-off points for high-stakes decisions; and continuous monitoring of model drift and bias. Firms that treat AI as a black box — even a very capable one — will struggle to satisfy the FCA's CASS rules, the SRA Code's competence requirement, or upcoming Lloyd's Blueprint Two standards for insurance. This is why governance-first products matter. Trovix Audit builds compliance visibility into AI workflows by design, not as an afterthought. It's not about slowing things down. It's about making speed defensible.

If you're a mid-market financial services firm, law practice or insurer, the practical action is simple: before you deploy any AI agent into regulatory-sensitive work, map the governance requirements — FCA rules, PRA expectations, SRA standards, your own risk appetite — and build that into your architecture. Don't chase the headlines about what FIS did in June. Build what you can defend in a regulatory conversation in 2027. That might mean starting with lower-risk use cases (intake, triage, first-pass document review) where governance feels less onerous, then expanding. It means choosing AI vendors and tools that give you visibility into how decisions are made, not just what the output is. And it means treating 'trust' as an engineering problem, not a marketing claim.

Source: Forbes