AIG CEO Eric Andersen has laid bare what mid-market UK insurers already know but struggle to articulate: AI infrastructure is generating coverage exposures at a scale and concentration that traditional risk assessment cannot keep pace with. Data centers demand protection across project finance, construction, operations, and supply chain—each phase introduces novel failure modes. For a mid-market insurer or broker operating under FCA Consumer Duty requirements (PS22/9), this means your clients are asking for products you cannot yet adequately price or monitor. The capacity squeeze is real. But it is also symptomatic of a deeper problem: most insurers are adopting AI tools—whether for claims triage, underwriting recommendation, or document review—without first building the governance infrastructure to understand what those tools are actually doing to their own risk profile.

This story reflects a pattern that should concern every regulated firm in insurance, legal services, and financial advice. The industry is treating AI as a faster way to do the same work. But AI fundamentally changes what work is being done and who is liable when it goes wrong. When Luminance or Harvey are used to accelerate due diligence or claims assessment, they make decisions at scale that were previously made by humans who could explain their reasoning. When those decisions drive underwriting positions or coverage denials, they create a new class of regulatory exposure—one the PRA has quietly begun flagging in stress testing frameworks (PRA SS1/23). AIG is not warning about data center risk. AIG is warning about the blind spot in how AI is being integrated into the business. UK regulators are watching. The ICO's guidance on AI and UK GDPR is tightening. Lloyd's Blueprint Two is pushing for transparency in algorithmic decision-making. Firms that have not yet implemented proper AI governance are now operating on borrowed time.

Here is Trovix's honest view: the problem is not that insurers and brokers are using AI. The problem is they are using it without first asking what they need to know about it. Products like Luminance excel at pattern recognition in documents, but they do not tell you whether the patterns they have learned are fair, legally defensible, or auditable under Lloyd's requirements. AI claims automation tools can reduce processing time by 40%, but if you cannot explain why a claim was recommended for denial, you have a Consumer Duty breach. This is where Trovix Audit comes in—not as a faster AI tool, but as a way to build the compliance and governance layer that makes other AI tools safe to use at scale. The difference matters. Firms using unmonitored AI without governance frameworks are essentially running blind; they are betting that their black-box tools will not become a regulatory liability. Trovix Audit is built specifically to address this gap—to give you continuous visibility into how your AI systems are performing against regulatory expectations, not just business KPIs.

If you are a mid-market insurer, law firm, or financial services provider, the action is clear and urgent. First: audit your current AI deployments—claims automation, underwriting support, document review, client intake—and document what each system is doing, how it is making decisions, and whether you can explain those decisions to the FCA or ICO if challenged. Second: implement a governance framework before adding more AI tools. This is not a 'nice to have'; it is now a compliance requirement under Consumer Duty and emerging PRA expectations. Third: when you do integrate new AI, build in explainability and audit trails from day one. Trovix Brief handles intake automation with full traceability; it is AI integration done with governance baked in, not bolted on after. The AIG warning is not a forecast of future risk. It is a statement of present capacity. UK firms that have not yet faced that capacity problem are closer to it than they think.

Source: Bloomberg News