Visa's six new AI tools for dispute management are designed to handle the 106 million charge disputes processed globally in 2025—a genuine operational pain point. For UK insurers, lenders and payment service providers, this matters because dispute resolution sits inside the regulatory perimeter. The FCA Consumer Duty (PS22/9) requires firms to design processes that deliver fair outcomes, not just faster ones. Automating dispute handling without explicit governance creates a compliance liability, not an efficiency gain. Visa's approach—using generative AI to resolve disputes before escalation—looks clean on paper. In practice, it shifts the burden of explainability onto firms using the tools, not the tools themselves.

This story is part of a broader pattern: major financial infrastructure players are embedding AI into processes that were previously human-reviewed, human-accountable and human-auditable. Payment networks, card schemes and settlement platforms are automating decisioning without acknowledging that their clients—mid-market acquiring banks, payment facilitators, insurance claims handlers—inherit the regulatory responsibility. The FRC, PRA and ICO are all circling this space. The EU AI Act is already here. Yet vendors are releasing dispute tools, claims tools and KYC tools as if governance is the client's problem, not the product's problem. It isn't. Under ISA UK standards, audit trails and explainability are baseline. If Visa's tools cannot produce them, firms that deploy them are creating audit risk.

Trovix's position is clear: automation in regulated processes must start with governance, not speed. Tools like Harvey and Luminance have built explainability into document review because legal risk depends on it. Visa's approach is backwards—it optimizes for merchant convenience and dispute prevention, which are good things, but it does not optimize for regulatory defensibility, which is essential. If a consumer challenges a dispute decision made by Visa's AI, and a firm cannot explain why the tool made that decision or how it was validated, the firm loses. Trovix Audit exists precisely because governance dashboards are non-negotiable in regulated AI deployment. Before any firm plugs Visa's (or any vendor's) dispute tools into their process, they need visibility into how those tools work, what data they use, what decisions they make, and whether those decisions are auditable. That infrastructure has to be in place first.

For mid-market law firms, insurers, and financial services firms: do not adopt Visa's dispute tools (or any similar product) based on operational efficiency alone. Audit your current dispute handling process against FCA Consumer Duty and PRA SS1/23 first. Identify where human judgment is currently required and whether automating that judgment creates a compliance gap. Then, if you move to automated dispute tools, implement governance infrastructure—audit logging, decision explainability, performance monitoring—before go-live. Trovix Audit is built for exactly this: tracking what AI systems decide, why they decide it, and whether they are performing fairly. The vendors selling you speed are not selling you compliance. Those are different products.

Source: CNBC