NEW YORK — Major insurance providers including AIG, Great American, and WR Berkley have petitioned U.S. regulation officials for permission to stop covering AI risk, admitting that computer code has finally become scarier than actual acts of God.

The industry giants, whose entire business model relies on betting money against disasters, argued that while they can predict the cost of a massive earthquake, they cannot predict what happens when a chatbot decides to lie to a customer. In filings, underwriters described the technology as a “black box” that defies the laws of probability.

“We understand how a tornado works,” said Bill Henderson, AIG’s Vice President of Betting on Bad Things. “Wind blows, houses fall down, we write a check. It’s simple math. But these computer programs are making things up faster than we can print money. I will insure a fireworks factory built inside an active volcano before I insure a predictive text generator.”

The move to exclude AI liabilities has confused business owners who assumed insurance existed specifically for risky things. However, experts say the math just doesn’t work when the software changes every week.

“I can calculate the odds of a meteor hitting your headquarters,” said Sarah Jenkins, Director of Panic Management at Great American. “But a computer program that hallucinates fake legal precedents? That is pure chaos. We are in the business of risk, not magic.”

At publishing time, AIG had updated its standard business policy to explicitly cover alien invasions, zombie outbreaks, and the collapse of the sun, noting that those events are “at least consistent.”

Inspired by actual events.