How AI-related risks can affect insurance

Most organizations have by now heard the warning bells of risks that come with artificial intelligence technology, from data privacy and cybersecurity threats to potential copyright infringement and discrimination claims.

In face of the spike in AI-related litigation over the past couple of months, such risks could soon prove costly for many organizations, leaving companies with one last barrier of defense: insurance.

While there is no such thing as an AI-specific insurance policy—yet—enterprises’ existing coverage may protect them against some of the litigation threats created by artificial intelligence. But as the risk for regulation and waves of lawsuits grow, the extent to which insurers will protect companies is uncertain.

During the “Insurance Considerations for Deploying AI/ML Tools and Technologies” webinar hosted by Clark Hill on Wednesday, legal professionals highlighted some of the key risks associated with artificial intelligence and the insurability—or lack thereof—of these risks under traditional coverage.

Does insurance cover AI risks?

While the majority of existing insurance policies don’t explicitly mention AI, legal professionals noted that many of the risks that companies face because of the technology are protected in some shape or form by their existing coverage.

Take, for example, the cybersecurity risks that come with using AI or, more recently, with generative AI. Most organizations likely have a cybersecurity policy that may protect them from some of the ransomware, data breach and other cyber liabilities. But the sophistication of cyber threats that generative AI can spur should bolster organizations to take another look at such policies.

“You really need to know what your vendors’ terms and conditions say and what your cyber policy covers,” said Jennifer Stivrins, a partner at Kissel Straton & Wilmer.

Still, cyber policies are only one part of the puzzle when it comes to protecting organizations against cyber risks, she said, as other types of policies may also include some limited cybersecurity coverage.

“If you’re getting a business policy, if you’re getting a [directors and officers] policy and they offer you a cyber endorsement, look at that endorsement, see what it covers, is it a little bit more far-reaching?” she said.

Still, as the cyber risks surrounding AI continue to grow, organizations without the proper security and governance strategies in place may have more trouble getting coverage. In fact, going forward, Stivrins noted that she expects to see carriers start asking for acceptable use policies, which define how employees can use an organization’s information technology resources, such as internal AI systems.

“I suspect we will get there. I think we are there already in some instances. We have seen on applications asking for lists of the vendors that a given entity might use, ‘Do any of them employ AI technology?’ Particularly in the media sphere,” she noted. “But I do think we’ll get to that point where we are having to show a policy of how we use AI.”

‘You want [vendors] to indemnify you’

Of course, cyber risks aren’t the only threats to organizations when it comes to AI. Intellectual property and copyright have been hot areas for litigation in the past couple of months, especially regarding the type of media used to train generative AI models.

This type of litigation risk extends beyond creative professionals and enterprises, Stivrins warned, and should be on every company’s radar.

“You need to be asking your vendors a couple of questions. Are you using AI tools or technology to deliver services to my company? And if so, what were those AI tools or technology trained on? And does that include any protected content?” she said.

Stivrins added, “Ideally, you want them to indemnify you for intellectual property infringement risks, because they’re the ones creating the imagery or the language. I would avoid any AI tools that can’t confirm or at least don’t assert that they use proper licensing of their source material.”

Here again, several business policies may cover some of these risks. A stand-alone media policy is one example, Stivrins said, “particularly if you have any place where you’re putting out content that includes a website.”

General liability policies also may have an advertising cover provision that extends to risks such as copyright infringement, she added.

Are lawyers’ mistakes insurable?

Of course, one of the biggest threats that has emerged in the face of the generative AI wave is to legal professionals themselves. One of the most infamous examples is perhaps the New York lawyer who submitted a ChatGPT-written brief with fake case citations to the court.

Here, too, it seems that kind of mistake can be insurable under current insurance provisions, as long as one’s liability policy doesn’t have any kind of exclusionary language in it for generative AI, noted Meredith Challender, a partner and general counsel at Kissel Straton.

But such exclusionary language could soon become more common.

“We are starting to see that in some policies, some insurers [are] trying to include that,” she said.

Ultimately, those that may have the trickiest time getting sufficient insurance coverage may be the providers of AI services and products themselves.

Up until now, technology errors and omissions insurance has been providing coverage to companies creating software for other businesses to protect them in the case of a mistake. But the extent of that coverage may shrink going forward as the number of risks that can fall into that basket continue to grow.

“We are going to see tech companies that produce AI seeking insurance for that kind of work. And I would assume that they will get it but it will be somewhat limited to what the policies are going to provide,” Challender said. “I can see a lot of insurance companies saying ‘OK, to a point, we will ensure that. But we’re not going to be there to cover every single thing that might eventually go wrong.’”

Related: Don’t fear that AI is here: How the emerging tech can help HR pros

Looking ahead, it’s likely that generative AI will bring its own wave of changes to the insurance industry. While there aren’t bespoke AI-specific insurance policies yet, that may change as the insurance industry becomes more knowledgeable about the technology and adapts its types of coverage provisions.