Skip to main content

To misquote a Chinese proverb,

一字千金  —  One word is worth 1,000 pieces of gold

this is certainly the case in highly regulated industries. In the insurance and law sectors, a single word (or even a punctuation mark) can cause a ripple. In the wake of the coronavirus outbreak, we see many interesting examples of how subtleties in policy wordings can cause substantial effects. 

In this blog post, we are aiming to shed some light in this regard to underline why LegalTech and InsureTech are fascinating but challenging fields to work in.

Legislation: “serious and imminent threat”

Drafted in response to the coronavirus outbreak, the UK government’s Health Protection Regulations (2020) is obligated to declare a “serious and imminent threat” in order to quarantine suspected virus patients. While the use of such words is certainly dramatic, it is likely needed for obscure legal reasons.

Interruption: “disease” vs “notifiable disease”

Insurance firms will not pay business interruption cover unless the government has declared coronavirus to be a “notifiable disease”. In this sense, a “disease” alone is not sufficient to trigger a claim, and only a “notifiable disease” is covered. The additional complexity is that many policies list specific diseases that are covered. As COVID-19 is a new disease, it is not included in most of the lists. Furthermore, standard business interruption cover does not include forced closure by authorities, and it is only available as an extension, which most small business would not have purchased prior to the outbreak.

Production: “advisory measures”

Production insurance would not be payable for theatre productions before the UK government announced a ban, as the previous measures were only advisory. Loss sustained prior to the official ban are non-indemnifiable. 

Cancellation: “pandemics and infectious diseases”

Insurance firms will not pay for a cancelled trip because “pandemics and infectious diseases” are not covered under some travel insurance policies. In some policies, “infectious diseases” are not covered, whilst some do not cover “epidemic and pandemic-related costs”. Generally, “infectious disease” is a much broader definition than “pandemics” – but underwriters can declare a disease to be a “pandemic” (prior to the official WHO declaration) and therefore voiding the coverage.

Travel: “coronavirus exclusion”

UK insurers Aviva and LV suspend travel cover due to unprecedented demands, and UK specialist travel insurer Sportscover Direct, introduced a “coronavirus exclusion”. These measures are designed to limit the liabilities and risks from the insurers’ perspective, but obviously they are not good news for the consumer. 


The examples above show an outline of subtleties and therefore challenges in policy wordings and the ripple effect of each word. Measuring small text changes that are not presented in a statistically significant way is always going to be challenging, and most NLP tools and models are not primed for identifying “obscure legal words”.  

Currently we have partnered with a leading global law firm to develop Natural Language Processing language models specifically designed for the insurance and legal sectors, as well as AI solutions that can help humans make better informed liability decisions. We have two ongoing projects that focus on using AI and NLP to:

  1. Quantify risk and analyse coverage of insurance policy wordings
  2. Augment and handle insurance claims liability decisions 

In order to measure subtleties that are otherwise difficult to be identified using automatic means, we have implemented language models which are drawn up in partnership with legal experts, and the algorithms are designed to extract and flag keyword changes which could affect coverage. 

This challenge is for our specialists a truly exciting opportunity, so watch this space for further updates!