When it comes to artificial intelligence liability, there are still so many unknowns, a Gallagher Re executive said during a panel at the PwC Insurance Summit on Wednesday.

Freddie Scarratt, insurtech global deputy head at the firm, said if a client sues because of bad AI advice, it is sometimes unclear whether the loss is covered under a cyber policy or a liability policy.

To that end, Gallagher Re has formed a task force to look at AI liability and pricing.

“We are trying to help customers and clients understand what can be done about it,” Mr Scarratt said. “Step one is to quantify the risk, and if you quantify it, are you pricing for it? It is a really interesting question, which is going to come to the fore in the next six months.”

He said one of the big questions is how to underwrite AI liability.

“Do you do it in a traditional way?” Mr Scarratt asked. “Do you look at the lawsuits? Do you try and get a frequency severity, or do you actually plug into people’s AI systems and stress-test them?”

The executive said that Gallagher Re’s retail side is seeing clients asking for AI liability coverage.

“They say they have just employed a chatbot to do a lot of the work for them,” Mr Scarratt said. “They are relying on that output and want to mitigate their risk.”

The panel’s moderator, Lee Harris, insurance correspondent at the Financial Times, asked Mr Scarratt if there are balance sheets big enough to face some of the potential claims.

She pointed to Anthropic, the large language model that last month was ordered to pay $1.5 billion to writers who alleged the company used their work to train its models.

It was the largest copyright recovery in American history.

“Is there insurance capacity to meet that kind of event?” she asked.

Mr Scarratt’s answer was: “Who knows? I would love to be the person who takes that risk to market.”

He said he had taken “one of these” start-ups to market and been pleased with the response from the Lloyd’s market, where the risk is placed.

“A lot of people have been very open about wanting to take us on their balance sheet,” he said. “We have raised all the capacity we need and then we will be launching it in January.”

Speaking about the impact of AI on various aspects of the insurance industry, Christina Lucas, global director and market leader, insurance, Google, said the focus is shifting from the back office to the front office.

“It is more about business value creation and less about efficiency,” Ms Lucas said.

Mr Scarratt said claims are the first area of the insurance industry to really see AI-driven advancement in areas such as fraud detection and customer experience at the consumer and small and medium-sized enterprise level.

“The area that is most exciting is the risk level,” he said. “The conversation is around, how do we free up the underwriter so that the underwriter can concentrate on writing complex risks? From a broker point of view, it is how do we free up the brokers to work on client solutions rather than the tedious, everyday order and management, for example? So in some of the areas, we’re seeing a lot of traction.”

He said the next step is using AI at the complex risk level.