What are the emerging risks reshaping Professional Indemnity and Financial Institutions insurance?
The risk landscape for professional services firms and financial institutions is changing remarkably quickly.
What I find particularly interesting is how technology, geopolitics, regulation, and new financial structures are changing how professional errors occur, how financial decisions are made and, ultimately, where liability sits. AI is perhaps the clearest example.
Professional services firms are increasingly incorporating AI into legal services, accounting, valuation and consulting. Professional errors may therefore arise not only directly from their errors and omissions, but also from inadequate verification of AI-generated output, poor data quality, loss of confidentiality or privilege, embedded bias and insufficient human oversight and control.
The (re)insurance market is already preparing for this. Recent Lloyd’s Market Association research found Professional Indemnity to have the highest potential impact among the AI loss scenarios examined. In a scenario where AI produces erroneous professional services causing client loss, 58% of responding underwriters considered that losses could potentially reach full policy limits.
The discussion is consequently moving from “Is AI a risk?” towards much more practical questions around wording adequacy, governance, underwriting and systemic exposure. Lloyd’s market discussions have also specifically considered legal professional privilege, public AI tools and the importance of human oversight and control.
From an underwriting perspective, this also means that simply asking whether a (re)insured “uses AI” is becoming increasingly insufficient. The more relevant questions are where it is used, whether it contributes to client-facing or material decisions, what level of human validation exists, which models and providers are relied upon, and whether the organization can identify and correct an erroneous output before it creates a loss.
Pricing will need to evolve with this. Historical claims experience remains important, but it may become less predictive where the underlying professional process has materially changed. Quality of controls, degree of automation and dependency on third-party models may increasingly need to influence pricing, retention and capacity decisions alongside traditional rating factors. For Financial Institutions, the exposure goes further.
AI is increasingly involved in decisions, investment management, fraud detection and risk management. Financial institutions and professional firms are increasingly relying on the same external technology, data and AI providers. A weakness in one widely used model or provider could therefore affect multiple (re)insureds simultaneously. This concentration also creates an aggregation question for (re)insurers. Two individually well-managed institutions may still represent correlated exposure if critical processes depend on the same model, technology provider or dataset. Understanding these common dependencies may therefore become increasingly relevant not only at individual risk level, but also for portfolio management and capacity deployment.
At the same time, synthetic identity, deepfakes and sophisticated social engineering are challenging traditional authentication. A fraudulent instruction may now involve a replicated executive voice, a convincing video call or an entirely synthetic customer.
Here, underwriting may need to move beyond traditional dual authorization and call-back procedures towards understanding how (re)insureds authenticate identity when voice, image and even live video can potentially be fabricated. Stronger controls should ultimately be capable of creating meaningful pricing differentiation rather than emerging exposures being addressed solely through exclusions.
From an underwriting perspective, what is particularly interesting here is the convergence between traditional Financial Lines products. Lloyd’s market discussions are already highlighting increasing AI, cyber and social engineering exposures and, importantly, the growing overlap between Crime and Professional Indemnity risks. Existing FI wordings are consequently being tested against scenarios they were never originally designed around.
Geopolitical fragmentation adds another layer.
For PI and FI, the exposure is not the physical conflict itself. Recent developments in the Middle East demonstrate how quickly counterparty risks, cross-border restrictions and regulatory requirements can change. For financial institutions, this increases transaction and compliance exposure. For lawyers, consultants and other professionals, it increases the possibility of professional liability arising from incorrect, incomplete or simply outdated output.
Private credit is another area I would watch closely.
The global private credit market is now estimated at approximately USD 1.5-2 trillion and is becoming increasingly interconnected with banks, (re)insurers, asset managers and private equity. Regulators are focusing more closely on leverage, valuation, borrower quality, liquidity, concentration and transparency. For FI underwriting, this makes transparency around underlying exposures, valuation methodology, concentration, leverage and conflicts increasingly relevant to both risk selection and pricing. Growth itself is not necessarily the problem; rapid growth without equivalent development of governance and controls is. Meanwhile, the boundaries of the financial sector itself continue to blur.
Fintechs, embedded finance providers and non-bank lenders increasingly perform activities once associated almost exclusively with traditional financial institutions. Traditional institutions, in turn, depend more heavily on third-party technology, data and AI.
This leaves us with an increasingly difficult question:
When a professional or financial decision is produced using a third-party model, platform or dataset, who ultimately owns the error?
Perhaps that is the real emerging risk.
Not AI, geopolitical instability, sophisticated fraud, private credit or regulatory complexity individually, but their convergence.
AI can create a professional error. The same technology can facilitate fraud. That fraud can reveal weaknesses in a financial institution’s controls. A regulatory investigation can follow. And the resulting loss may sit somewhere between PI, Crime and other Financial Institutions coverages.
This convergence may ultimately require a more dynamic underwriting approach: less reliance on static questionnaires and historical classifications, and greater focus on how the (re)insured actually operates, how quickly its risk profile is changing, where critical dependencies sit and whether pricing, retention, limits and wording remain proportionate to that exposure.
For me, this changes the underwriting question.
It is no longer only:
“What new risks are emerging?”
But also:
“Do our existing controls, underwriting assumptions and policy wordings still reflect how professional and financial liability is actually being created today?”
The products may still be called Professional Indemnity and Financial Institutions insurance. The risks inside them are changing much faster.
Comments