People whose job it is to underwrite risks say AI is too risky

What happens when the software everyone is racing to adopt becomes too risky for anyone to secure? According to a report from the Financial Times, we are about to find out.

Major insurers, including AIG, Great American, and WR Berkley, are asking US regulators for permission to exclude AI-related liabilities from corporate policies. One guarantor describes the output of the AI ​​models to the Financial Times as “like a black box.”

The story reminds us that the industry has good reason to be afraid. Google AI Overview falsely accused a solar company of causing legal trouble, leading to a $110 million lawsuit in March. Last year, Air Canada stumbled to respect an opponent invented by its chatbot. Last year, scammers used a digitally cloned version of a senior executive to steal $25 million from London-based design engineering firm Arup during a video call that appeared completely real.

What really scares insurance companies isn’t huge payouts; It’s the systemic risk of thousands of simultaneous claims when a widely used AI model comes into play. As one Aon executive said, insurers can handle a $400 million loss for a single company. What they can’t handle is an AI mishap that causes them to lose 10,000 at once.