Chat, am I covered?

If a restaurant burns down, property insurance may cover the damage. If a customer slips and falls, liability insurance may cover the lawsuit. But what happens when an OpenAI employee hides a recording device wrapped in toilet paper inside a conference room?

While it’s an unusual case of corporate espionage, the incident underlines a broader question the insurance industry is asking: Who pays when things go wrong in the AI business—and should insurers get in?

Chatbots can give dangerous medical advice, autonomous agents can leak confidential information, and AI-generated content can trigger copyright lawsuits, to start. And with risks still growing, traditional insurance policies aren’t designed to address them.

Enter silent AI risk: According to broker Aon, more than 90% of insured AI issues fall into this category, meaning existing policies don’t explicitly include or exclude them. In a week where major insurers reported earnings, executives had the chance to take a position.

A premium opportunity

Whether AI becomes an insuring gold mine or a financial disaster remains an open question. The uncertainty has already prompted major insurers like AIG, Berkshire Hathaway, and Great American to seek approval for excluding certain AI-related claims from their policies. But others are addressing the potential head-on.

The upside

As AI adoption expands, insurers could collect premiums from an entirely new market. Munich Re offers specialized AI insurance with coverage limits reaching $50 million, while specialty insurers affiliated with Lloyd’s of London are developing similar policies.

Marsh CEO John Doyle warned in Thursday’s earnings call that AI is already “embedded in most companies’ workflows,” making blanket exclusions a source of “real confusion and uncertainty.” The insurance broker argues that existing policies need to evolve alongside the technology rather than simply exclude it.

The downside

Nobody knows exactly how much these risks are worth. Unlike car accidents or house fires, AI failures have little historical precedent to help insurers predict their costs. Plus, when businesses rely on the same AI systems, a single malfunction could trigger claims across many policies, leaving insurers paying out far more than they collected in premiums.

That was the verdict of Travelers executives, who acknowledged that scope creep during Friday’s investor call. To rein the risk in, the company said it limits most of its cyber policy coverage at $1 million and relies heavily on reinsurance to limit potential losses.

Artificial liability

Yet in a week packed with industry earnings and investor calls, AI liability coverage received surprisingly little attention. The relative silence suggests that, for all its potential, insurers have yet to figure out how to turn AI’s growing risks into a meaningful source of profit. For investors looking for the next winner of the AI boom, insurance may not yet be a clear bet. But should the robots ever stage a rebellion, let’s hope someone remembered to purchase the optional apocalypse coverage.—SY