26 April 2023

US insurers block and moderate access to ChatGPT

Reinsurance Group of America (RGA) has temporarily blocked general access to artificial intelligence (AI) chatbot ChatGPT, the reinsurer’s global chief risk officer (CRO) Jonathan Porter has said. 

During a CRO roundtable organised by the North America CRO Council yesterday, panellists discussed the potential risks arising from using data analytics and AI in insurance risk management, including the use of ChatGPT.

Released in November 2022, ChatGPT is a natural language processing tool powered by AI technology that provides “human-like” text responses to a given prompt. How it processes information has raised concerns over privacy and personal data usage. At the end of March, the Italian Data Protection Authority banned ChatGPT following a data breach.

Roundtable moderator Porter said RGA understood the need to experiment with AI tools, including ChatGPT.

“But we've actually temporarily, or at least at this point, blocked general access to ChatGPT and other tools within our company just to protect us from issues about privacy concerns and wrong answer concerns,” he said.

He asked other panellists if they had taken similar measures.  

“We have actually not blocked company access to ChatGPT,” said Rahim Hirji, CRO at Canadian life insurer Manulife.

“We actually did a risk assessment where our view was that given there is a limited amount of information you can actually feed it using the public aspect, we were comfortable with taking that risk.

“We will monitor it and we are sort of monitoring volumes to see what is actually being fed [to it] in terms of data.”

Klaus Diem, CRO at US insurer Nationwide, and Deb Shultz, CRO at Northwestern Mutual, both added they were trying to limit usage of the tool while still experimenting with it.

“What I mean by that [trying to limit usage] is we're trying to monitor who has access to make sure that we can understand what they're trying to do with ChatGPT,” said Diem.

“As this technology is evolving and as we better understand it, I think we have to educate our associates in terms of what it means in confidentiality and putting things out there.”

Shultz said Northwestern Mutual is adopting a similar policy and taking a “very measured approach” to the use of ChatGPT and other related tools.

“Obviously, this is moving very fast and there's a lot to sort through, so we need to manage it from a company risk perspective, but also recognise some great opportunities with [the use of] advanced analytics and the ability to drive efficiencies,” Shultz added.