SIRC 2026: AI will ‘go all the way’ to autonomous underwriting, Scor’s Thierry Léger predicts

November 2 2026 by

Scor CEO Thierry Léger effectively signalled the end for many human underwriting roles as he explained how AI will transform the market in an address at this year’s Singapore International Reinsurance Conference.

“I think AI is exciting … I never thought I would see something so transformative for our industry,” he said on Monday.

“We all know that are at very much at the beginning of this in our industry. It’s very clear that this is an opportunity for us. It’s an opportunity on the business side, but it’s also an opportunity to transform the way we do reinsurance, our business models.”

He added that we should clearly not put barriers in its development, but “we have to put in place clear guardrails to support the development of AI in our companies” as “we are very much entering – and have probably entered already – a world in which we have a hybrid workforce”.

Léger said he was “convinced” that we should handle AI agents in the same way that we handle a human employee.

“We have to find a way to register every agent that is doing work, and certainly at Scor we have no intention that any agent that is not properly registered is doing work for our company,” he added.

He said there will be a need for enabling oversight, adding that Scor has four levels of governance, with level three when AI is able to do all of the work of an underwriter, and can compose a quote to an underwriter.

Crucially, however, the human underwriter is making the ultimate call and sending the offer to the client.

Level four, he added, is AI making all the decisions itself and sending the offer to the client autonomously.

“In my view that’s a big step,” he said. “We want that step, from three to four, to be controlled. We are not trying to block anything, we just want to know which agent is actually allowed … to bind externally.

“Of course this will have a massive impact on our workforce. Everyone will have to embrace AI, and I think that’s exciting.

“For leaders, however, that’s going to be quite challenging.”

“It would in theory be possible to do [level four]. I’m pretty sure that at some point this will go all the way to level four: autonomous sending out of offers.” Thierry Léger, Scor

He suggested that the (re)insurance industry leadership really has no idea what the real differences will be in terms of managing humans and managing AI agents.

“We know that human beings have their strengths and weaknesses, but the agents we don’t know yet what their strengths and weaknesses are,” Léger said.

He alluded to the fact that, at present, AI can be considered less than 100% reliable, as given a dataset five times, the AI model would give five “slightly different answers”.

He also said that AI will not necessarily remember how it arrived at the answers, so that if the result is wrong, you cannot question it in the same manner that one could a human being.

As such, there is a need for clear protocol within the market to ensure that AI agents can log how they reached their decisions.

Léger told delegates that Scor uses AI a great deal at present, especially for contract management, and has also recently developed a pilot for augmented underwriting, which is able to ingest all the data, look at accumulations, and make some commercial decisions around pricing.

“It’s a very complete set of capabilities,” he added.

“It would in theory be possible to do [level four]. I’m pretty sure that at some point this will go all the way to level four: autonomous sending out of offers.

“But right now, we have stopped it at level three, and we want to learn from it and deploy it to many more lines of business.”

Partner Content
Join the mailing list Designed to help you stay ahead, our daily newsletter provides a round-up of the latest news and insights, delivered straight to your inbox.
Print Edition
Subscribe to receive our quarterly print magazine.