Remote to Insureds' Use of AI: A Lloyd's Market Association Survey of Underwriters on AI Loss Scenarios, January 2026

On January 22, 2026, the Lloyd's Market Association, which represents the interests of Lloyd's managing agents and members' agents, announced the results of a survey of its members, "Understanding AI Exposures: AI Loss Scenarios Survey Results." It is a market opinion survey, run "in the absence of market underwriting and claims data on AI exposures." It asked whether insureds are using AI now in ways that could cause or contribute to claims, which loss scenarios members are most concerned about, whether insureds are managing the risks adequately, and how large the losses could be. Of the 144 responses, collected in the second and third quarters of 2025, 94% came from underwriters.

The association drew up 65 loss scenarios across 10 classes of business. Not enough responses came in to analyze every one, it says, so its summary covers the four scenarios with the highest response rate, with a minimum of six responses each, though up to 19 responses were received for some scenarios. One of the four is professional indemnity.

Erroneous Advice to Clients

The professional indemnity scenario reads, "AI produces erroneous advice/service to clients, causing a loss." It was deliberately broadly drafted, the report says, and intended to capture views on circumstances such as incorrect tax advice given by an accountancy firm "due to a mistake made by AI software in analysing financial data," negligent legal advice from an AI program "hallucinating" non-existent case law, and architects' blueprints with material errors from faulty AI research.

In this scenario, 100% of responses suggested insureds were using AI at least to some extent, though just under half felt it was used "in a few cases" only. On the survey's scales, based on the responses, the scenario was rated "plausible" for viability and "medium" for magnitude, with an overall impact of "moderate." Of the four scenarios in the summary, it received the highest scores for both magnitude and overall impact.

The potential losses were estimated as reasonably significant. 58% of responses on this scenario thought losses might reach full policy limits. At the same time, 89% of respondents to the scenario felt that insureds' existing risk management was "adequate." This, the report says, "indicates a high degree of confidence in insureds to manage the risks arising from use of AI systems."

Junior Staff and Human Supervisors

"The risks envisaged in this type of scenario are not necessarily being considered as a new exposure by LMA members," the report says in its commentary on the scenario. "Reliance on AI may be analogous to reliance on (for example) junior staff producing tax calculations, legal research or technical drawings. The difference with AI might be a matter of scale."

The key function of AI, it adds, is "to use computing speed and power to let the system do the work and overcome human limitations," and "the benefits cannot be fully realised if the output must be constantly overseen by human supervisors."

Limits, Coverage and the Model Wordings

Respondents found the survey difficult to complete, the report says, for two main reasons. "Insurers are remote to insureds' use of AI," and little data on AI usage is currently collected during underwriting. And "there is a lack of claims data on AI contribution to losses." The responses, it says, are informed by views and opinions rather than data "in most cases," and the ratings "are not based on actuarial calculations or driven by claims, exposure or risk management data."

When estimating loss potential, respondents were asked to assume the loss was covered, and some high-level questions on the use of exclusions were included in the survey but are not reported in the summary. "The observations reported in this summary do not relate to specific losses, wordings or exclusions," the report says. Nor did the survey explore the protective effects of AI, which it says "could be material."

On wordings, the report draws on a related review, carried out in 2024, of the association's model cyber clauses. In that review AI is recognized as a sub-set of software, falling within the definition of "Computer System" used widely in those clauses, and an annex rates each clause's potential AI coverage as low, medium or high. That table, it says, "is intended to be indicative of potential AI (cyber) risk; the actual risks present in any contract will depend on the wording(s) used and the circumstances."

The Report's Conclusions

The risks of the scenarios it reviewed attracted, on the whole, "only a modest degree of concern amongst underwriters," the report concludes, and "were widely expected to grow in the future." Testing and risk management by insureds, it says, is key to ensuring the risks are managed to an acceptable level.

"The survey attracted a wide range of views and there is a lack of underwriting and claims data about AI exposures," it says. "Case studies, with detailed event profiles, applying specified wordings and limits of cover may usefully reveal more information about potential AI exposures."