Key Points
- Major insurance companies are seeking to exclude AI-related liabilities from their corporate policies.
- Insurers are concerned about the unpredictable nature of AI, which they describe as a “black box.”
- High-profile incidents, such as AI-related lawsuits and fraud, have spooked the industry.
- The biggest fear is the “systemic risk” of a single AI error causing thousands of simultaneous claims.
What happens when the new software everyone is rushing to use becomes too risky to insure? According to a Financial Times report, we are about to see just that with artificial intelligence.
Major insurance companies, including AIG, Great American, and WR Berkley, are seeking U.S. regulators’ permission to exclude AI-related liabilities from their corporate insurance policies. One underwriter described the outputs of AI models to the FT as “too much of a black box,” meaning they are unpredictable and hard to understand.
The insurance industry has good reason to be worried. Google’s AI Overview falsely accused a solar company of legal problems, leading to a $110 million lawsuit in March. Last year, Air Canada was forced to honor a discount that its chatbot simply made up. And in a more sinister case, fraudsters used a digitally cloned version of a senior executive to steal $25 million from a London engineering firm during a video call that appeared completely real.
What really scares insurers isn’t the possibility of a single, large payout. It’s the “systemic risk” of thousands of claims happening all at once if a widely used AI model makes a major mistake.
As one executive from the insurance broker Aon explained, insurers can handle a single $400 million loss to one company. What they can’t handle is a single AI mishap that results in 10,000 losses at once.
This move by insurers could have a significant impact on businesses’ adoption of AI. If companies can’t get insurance to cover the risks associated with AI use, they may be much more hesitant to implement the technology, which could slow the entire AI boom.