Companies Using AI Urged to Revisit Their Insurance Policies

According to an article in the University of California’s Continuing Education of the Bar (CEB) newsletter, brokers and attorneys say companies that use, plan to use, develop or sell AI will find that maintaining adequate coverage is a moving target. In light of recent safety incidents caused by rogue AI agents this summer, questions about what constitutes “adequate coverage” are becoming all the more urgent. Some businesses, the CEB report added, will discover they need entirely new coverage that before last year didn’t exist. And others will need to negotiate new terms that carriers are pushing into traditional policies.

In particular, new AI exclusion endorsements now exist, though they have been met with widespread pushback, said Pillsbury Insurance Recovery & Advisory lawyers Hadhy Ayaz and Tamara Bruno, who asserted that the changes went too far because a claims adjuster could argue that almost any machine-automated activity could be excluded.  “We said to the insurance companies: ‘We will not place policies with your insurance company… if you insist on including these exclusions,’” Bruno said.

Since then, Bruno added, insurers have pivoted, leveraging the exclusions as an underwriting threat to gain details about a company’s AI use. Insurers will say, “If you can't give it to us, we're going to include this exclusion,” she explained.

Unlike cyber liability, AI liability typically touches multiple parts of an organization, making it more challenging to assess. Other than for AI developers, Bruno explained, it’s difficult to isolate AI risk. “I think that's what insurance companies are having a really hard time with right now – they don't know how to separate AI risk from everything else because it's so enmeshed.”

Assessing one’s coverage is particularly challenging as AI is now being incorporated into so many business activities. “You have to understand how your organization uses AI. And I know how hard that is,” Bruno said. Pillsbury’s Ayaz added that circulating internal employee questionnaires can reveal to management how a workforce is using AI.

The profusion of AI lawsuits potentially could implicate lines like D&O coverage, Ayaz said. “I think that would be a space to be very cognizant of for additional exclusions and limitations on coverage as a policyholder.”

That said, he notes that some of his clients have avoided AI-based D&O carve-outs. “So, it's not like there's truly standard versions of these programs. Even ISO form endorsements can be tossed aside,” Ayaz said. “What that says is, there's space to negotiate.”

To learn more about AI and AI exclusions in insurance policies, see Pillsbury’s Policyholder Pulse blog here.