AI Insurance: Why Insurers Are Hesitant

The Growing Insurance Concerns Surrounding Artificial Intelligence

A critical question is emerging regarding the rapid adoption of artificial intelligence: what will occur when the inherent risks associated with this technology become too substantial for insurance coverage?

Recent reports from the Financial Times indicate that this scenario may be imminent. Leading insurance companies, including Great American, Chubb, and W. R. Berkley, are currently requesting authorization from U.S. regulatory bodies to exclude liabilities stemming from AI-related issues from standard corporate insurance policies.

One underwriter articulated to the FT that the outputs generated by many AI models are essentially “too much of a black box,” meaning their decision-making processes are opaque and difficult to assess.

AIG, also mentioned in the Financial Times article, has provided TechCrunch with a statement clarifying their position. They assert that they “were not specifically seeking to use these [reported upon] exclusions and have no plans to implement them at this time.”

Examples of AI-Related Incidents

The insurance industry’s apprehension is well-founded, as highlighted by several recent incidents. In March, Google’s AI Overview incorrectly alleged legal issues concerning a solar company, leading to a $110 million lawsuit.

Furthermore, Air Canada was compelled to honor a discount fabricated by its own chatbot last year. Fraudsters also exploited a digitally replicated senior executive to defraud Arup, a London-based design engineering firm, of $25 million during a convincingly realistic video conference.

Systemic Risk and Insurability

The primary concern for insurers isn't a single, large claim payout. Instead, it’s the potential for systemic risk – the possibility of numerous simultaneous claims arising from a failure of a broadly implemented AI model.

As an executive from Aon explained, insurers are equipped to manage a $400 million loss affecting a single entity. However, they are not prepared to handle a situation where an autonomous AI error triggers 10,000 separate losses concurrently.

This highlights the need for careful consideration of the risks associated with increasingly agentic AI and its implications for the insurance landscape.