BURLINGTON, Vt. — Captive owners and risk managers should embrace AI tools and predictive analytics as they look to mitigate the effect of nuclear verdicts on their insurance programs, experts say.
As jury awards exceeding $10 million become more frequent and severe, such technologies can identify high-risk claims early on and provide strategies to curb rising legal payouts.
Communication and early intervention are key to preventing large verdicts, the experts said during a panel session at the 40th annual Vermont Captive Insurance Association conference last week.
Actively communicating with actuaries, reinsurers and other stakeholders about emerging casualty claims is essential, said Melissa Hollingsworth, deputy chief risk officer of the Los Angeles Unified School District.
“You can’t sweep it under the rug. You have to be proactive and make sure you get it on people’s radars. I’m a big fan of communication. It goes a long way with helping mitigate these types of losses,” Ms. Hollingsworth said.
Reserving is crucial, she said. “Working with your actuary to predict and project out potential losses that are incurred but not reported, that IBNR, that’s really going to help, particularly with mitigating the impact on your captive,” she said.
In the first seven months of 2025, there were 70 verdicts of $10 million or more in the U.S., with an average payout of $238 million, said Denise Tyson, La Jolla, California-based CEO of Schaefercity.ai, an insurtech that provides AI products meant to help guard against such verdicts.
The 70 nuclear verdicts “impacted 20 states, so everybody’s at risk of these things,” she said.
It’s not just medical malpractice or sexual abuse and molestation. “Every casualty line is getting hit,” Ms. Tyson said.
Preventing claims from reaching trial is crucial because “once it goes to trial, as an insurer, as the claims person, it’s out of your hands. You’ve given it to a lawyer,” she said.
Technology can help by spotting red flags and guiding organizations on whether to settle a claim quickly or defend it, Ms. Tyson said. “There are tools out there, and we have to embrace these tools,” she said.
It’s important to stay open-minded and explore how different technologies can enhance and improve your insurance program, Ms. Hollingsworth said.
“Control what you can and use tools to help you build the best program possible,” she said.
AI tools do the grunt work, Ms. Tyson said. “All you have to do is analyze it and look at it and think about it differently,” she said.
Schaefercity.ai assesses the risk of a nuclear verdict by providing a prediction score. If a claim doesn’t settle, it offers claims professionals strategies to prevent a large verdict.
A risk retention group client used the scoring system to identify “red” cases — where the probability of a nuclear verdict is considered critical or extreme — so it could proactively fight the case instead of settling early, Ms. Tyson said. “She was able to settle it for less than half of the policy limit.”
A key challenge in using AI in claims is navigating ethical concerns, particularly around bias, said Derek Freihaut, Bloomington, Illinois-based principal and consulting actuary at Pinnacle Actuarial Resources.
“States are pushing back a lot on any type of bias that shows up in an actuary’s rating, when actuaries develop premiums,” and whether that unfairly affects certain groups, for example, said Mr. Freihaut, who moderated the session.
“AI is based on data and if the data already has underlying bias in it, that’s the input to the AI output,” but programmers and data scientists can create prompts to reduce bias, Ms. Tyson said.
“When we’re using these tools, you cannot take the human element out of what you’re getting from the data,” Ms. Hollingsworth said. “You still have to be heavily engaged and involved and cross your t’s and dot your i’s.”
Media Contacts:
Jordann Reis, Business Development Director
info@schaefercity.ai, (858) 257-0415
