Legal blind spots

Laurent Montador, Jérôme Isenbart and Aurelien Couloumy of CCR Group explore the legal blind spots that have materialised as a result of Covid-19 and how AI is assisting.

Covid-19 was a real shock for the whole world. Almost all industries have been impacted and the reinsurance market unfortunately did not escape the rule. Behind the sanitarian and economic drama, the context has also been the occasion of a serious introspection. It has put forward a series of questions for insurance actors, including two very interesting ones: “Can we improve the vision we have on what we underwrite and if not, how should we overcome the problem?”

When the disease and confinement started to progress, the whole market asked itself about specific details mentioned in reinsurance treaties regarding pandemic or business interruption expositions. More specifically, the market has been looking for very precise information about what was written into exclusions terms, business, and other coverage areas. Such interrogations could appear as trivial under the hypothesis of being equipped with risk management tools that contain data that truly illustrate contracts, which are otherwise standardized and are comprehensive contracts.

The problem is that reinsurance activity is by nature something that you cannot standardize because of the diversity of the activities, but also because of the business negotiations between experts that are used to bring tailor made coverages.

The disparity of reinsurance legal terms reflects this aspect well. Therefore, when it comes to studying the impact of an unprecedented event such as Covid-19, any particularities can become a legal nightmare to analyse. For instance, checking expositions consistently in contracts when on the one side there are general effective clauses that also demonstrate a lack of precision on certain terms, and on the other side, overdetailed clauses that provide very specific, but a non-exhaustive list of exclusions. It is complicated.

At CCR, we have been immediately very concerned by Covid-19. Following this complex situation, the company has set various impactful propositions both external (for example, insurance credit financial guarantees) and internal. Among the internal solutions, a range of innovative AI techniques have been used to overcome the potential of legal blind spot mentioned earlier and have received particular attention.

“At CCR Group, we have worked a lot over the last two years to find ways to tackle the crucial aspects oftextual unstructured data,” explains Laurent Montador, deputy CEO of CCR Group.

“We have created AI services that use natural language processing (NLP) that allow us to manage, among other aspects, legal terms issues efficiently. NLP has been particularly helpful for Covid-19 crisis management,” he added.

According to Jérôme Isenbart, CRO of CCR Group, “It mainly served three main purposes: to identify, manage and control textual data”.

“As a first step, NLP was helpful to identify precisely on a more automated fashion what wording we have in our contracts that is helping us to classify by topic specific clauses that could expose us,” he said.

He continued: “We also observed that it was sometimes quite hard to get a precise structured information overview. So, to get the most value from our internal knowledge, we have also developed a semantic search engine that uses a state-of-the-art language model called BERT (Bidirectional Encoder Representations from Transformers)”.

But the experience has not stopped there, as revealed by Aurelien Couloumy, head of digital transformation and also an actuary. “This Covid-19 context confirms our strong expertise of legal terms management, but also our thoughts that legal blind spots are a real risk in itself and that we had to continue our research to control and mitigate it.”

He concluded: “To do so, we have developed programs that are aimed at identifying outlier terms and sentences into reinsurance wordings and clauses. It also helps us investigate potential ambiguity which often appears as a problem and to avoid pitfalls of litigation.”