Professional Liability Insurance – Is AI Risk Insurable?

The AI Insurance Illusion: Why Securities Attorneys May Be Less Protected Than They Think

By Yan Ross JD, Editor-in-Chief, Cyber Defense Magazine

 

NOTE: The author is writing in his individual capacity only, and does not intend this blog to constitute legal advice or a legal opinion. Some research on the topic employed artificial intelligence resources, but the work product is the author’s alone.

 

The AI Insurance Illusion: Are Securities Attorneys Really Covered?

 

The greatest AI insurance risk facing securities attorneys may not be the absence of insurance. It may be something more subtle—and potentially more dangerous: believing the firm has AI coverage when the policy covers the technology but not the professional consequences of relying on it.

 

That distinction matters as generative artificial intelligence moves rapidly into everyday legal practice. Lawyers are already using AI for legal research, due diligence, SEC filings, disclosure review, regulatory analysis and client advice. The efficiencies can be substantial. So can the risks.

 

Consider a simple example.

 

A securities attorney asks the firm’s approved AI platform to determine whether a proposed transaction qualifies for an exemption from securities registration. The AI produces an impressive memorandum, complete with statutes, cases and citations. It looks authoritative.

 

There is just one problem. One of the cases doesn’t exist. Another exists but does not stand for the proposition the AI attributes to it. The lawyer doesn’t catch either error. The client relies on the advice and proceeds with the transaction. Investors lose money. The SEC starts asking questions.

 

What kind of loss is this?

An AI loss? A technology loss? A securities claim? Or plain old-fashioned legal malpractice?

 

The answer may determine whether there is insurance coverage at all.

 

We Have Seen This Movie Before

 

Professional liability insurance historically followed a fairly straightforward progression:

 

Professional service → professional error → client loss → malpractice claim.

 

Then computers and the internet arrived, and the picture became considerably more complicated. Traditional Lawyers Professional Liability (LPL) policies were never designed to deal with ransomware, massive data breaches, forensic investigations, network interruption or the cost of notifying thousands of people that their personal information had been compromised.

 

Cyber insurance emerged to fill those gaps.

 

Now AI is creating the next insurance transition. But there is an important difference between cyber risk and AI risk.

 

An AI system does not have to malfunction to cause a catastrophic loss.

 

It can work exactly as designed and simply give the lawyer the wrong answer.

 

For attorneys, that distinction is critical.

 

The ABA has made clear that lawyers using generative AI retain their traditional professional obligations, including competence, confidentiality, supervision and candor. AI may perform some of the work, but responsibility for the resulting legal work stays exactly where it has always been—with the lawyer.

 

“The AI made the mistake” is unlikely to satisfy the client, the court, the SEC or the malpractice carrier.

 

The Problem of “Silent AI”

 

Most professional liability policies in force today were originally designed before generative AI became part of routine legal practice. Many neither expressly cover nor expressly exclude its use.

 

That creates what the insurance industry increasingly calls “silent AI.”

 

Suppose a Lawyers Professional Liability (LPL) policy covers wrongful acts committed while rendering professional legal services. There is a strong argument that using AI does not change the nature of the underlying service. AI is simply another tool.

 

Lawyers have always relied on associates, paralegals, treatises, Westlaw and Lexis, databases, calculators, spreadsheets and document-review software. Why should using AI transform legal advice into something other than a professional legal service?

 

It shouldn’t necessarily.

 

But insurance coverage is determined by policy language, not common sense.

 

An insurer confronted with a multimillion-dollar claim may argue that the loss arose from technology services, data processing, investment advice, securities activity or some other excluded category.

 

The apparent answer is simple: buy an AI endorsement.

 

Unfortunately, that may be where the next problem begins.

 

An “AI Endorsement” Is Not Necessarily AI Malpractice Coverage

 

The word endorsement sounds reassuring. It shouldn’t.

 

An endorsement can broaden coverage. It can also restrict it.

 

Some insurers are introducing affirmative AI coverage. Others are introducing AI exclusions. And even an endorsement that affirmatively covers AI must be examined to determine exactly what kind of AI loss it covers.

 

A cyber insurer, for example, might cover an AI-created security failure, deepfake impersonation or fraudulent funds-transfer instruction. Those are real risks and valuable coverages.

 

But change the facts.

 

Suppose the AI simply gives the securities lawyer the wrong legal answer.

 

• There is no hacker.
• No ransomware.
• No unauthorized system access.
• No deepfake.
• No data breach.

 

There is simply bad legal advice produced with the assistance of AI.

 

A cyber policy carrying an impressive-sounding AI endorsement may provide little or no protection.

 

The claim goes back to the lawyer’s LPL policy.

 

And that is where securities attorneys encounter another problem.

 

The Securities-Law Trap

 

Securities practice occupies an unusually complicated corner of professional liability insurance because some LPL policies contain exclusions involving securities activities, investment advice, regulatory proceedings, governmental claims, fraud, fines or penalties.

 

Now follow the chain:

 

AI error → lawyer relies on it → inaccurate disclosure → investor loss → SEC investigation → civil litigation.

 

Which policy responds?

 

Perhaps several—and perhaps none completely.

 

Cyber insurance may respond if confidential information was compromised. An AI endorsement may respond to a specified technology failure. LPL may cover the lawyer’s negligence. D&O insurance may become relevant to management-related claims. Crime insurance may respond to an AI-enabled fraudulent transfer.

 

But suppose the LPL policy contains a securities or investment-advice exclusion broad enough to reach the underlying transaction.

 

The firm may have purchased excellent AI coverage while failing to insure the professional activity that actually produced the loss.

 

The AI endorsement becomes an umbrella over a house with no roof.

 

Stop Asking, “Do We Have AI Coverage?”

 

That is the wrong question.

 

A broker may truthfully answer “yes,” and the answer may tell the securities firm almost nothing about its actual protection.

 

Instead, give the broker a realistic hypothetical:

 

One of our attorneys uses an approved generative-AI system while providing securities legal services. The AI produces erroneous information. The attorney fails to catch the error. The information enters a securities filing, legal opinion or client recommendation. Investors lose money. The SEC investigates. The client and investors sue the firm.

 

Then ask the questions that matter:

 

Who pays for our defense? Who pays the damages? Who pays for the SEC investigation? Which exclusions apply? What sublimits apply? And, when all the policies have responded, what remains uninsured?

 

Then ask for the answer in writing.

 

That exercise will tell the firm considerably more than asking whether an insurance policy contains the words “artificial intelligence.”

 

Insure the Consequences, Not Just the Technology

 

AI risk is increasingly insurable. Specialized insurance products are beginning to address hallucinations, model failures, intellectual-property claims, privacy losses, deepfakes and AI-enabled cyber events.

 

That is encouraging.

 

But insuring the failure of an AI system is not the same thing as insuring the lawyer who relies on it.

 

For securities firms, meaningful protection increasingly requires coordination among LPL, Cyber and Affirmative AI Coverage, supplemented where appropriate by D&O and crime insurance. Just as important, the policies must be examined together for securities exclusions, investment-advice exclusions, regulatory coverage gaps, sublimits and conflicting definitions.

 

The central lesson is remarkably simple:

 

Don’t merely insure the AI. Insure the consequences of using it.

 

The most dangerous insurance gap of the AI era may not be discovering that your firm has no AI insurance.

 

It may be discovering—after the SEC investigation has begun, the client has retained litigation counsel and the malpractice complaint has arrived—that the firm did, indeed, buy something called “AI coverage.”

 

It may turn out that it just didn’t cover the AI risk that mattered.

 

###

 

You can reach the author for additional information by email at yan.ross@cyberdefensemagazine.com