A.I. Did What? — Don’t Let Your Insurance Company Use the Involvement of A.I. Technology to Reduce Your Insurance for a Claim

Policyholder Advisor & Alert

PUBLISHED ON: June 7, 2023

Download PDF

Key Points:
  • A lawyer recently filed a brief citing several cases that it turned out had been invented (unbeknownst to the lawyer) by ChatGPT.
  • How might insurance coverage for liability stemming from such an error play out? Using un-artificial intelligence, we map out likely scenarios.
  • For novel claims, expect – and be prepared to fight – novel coverage defenses.


Artificial Intelligence Artificial intelligence tools permit humans to make bigger and more grandiose mistakes.  Insurance companies make things even worse with improper claims practices.

A New York Times account of a lawyer who used ChatGPT to prepare a court filing shows how reliance upon A.I. tools can multiply the impact of errors.

In an affidavit filed in Federal Court in New York, the hapless lawyer declares, “It was in consultation with the generative artificial intelligence website Chat GPT (sic), that your affiant did locate and cite the following cases in the affirmation in opposition submitted, which this Court has found to be nonexistent:” (emphasis added)

In other words, ChatGPT appears to have invented six (6) nonexistent cases for this lawyer to rely upon in a court filing. Also, the same affidavit shows the poor lawyer also asked ChatGPT to confirm the existence of the imaginary cases, and it did so.

Thus, this lawyer has made a substantial mess for himself, his colleagues, and his law firm. Without the disarming power of A.I., the lawyer would have avoided the mistake.

Insurance Companies Take Advantage

To create the case history in question, ChatGPT drew various scraps of information from its unfathomably large database and assembled an alternate reality. As possessors of organic intelligence, we can do that too. While there is no public record about any insurance claim the hapless lawyer in this case may have filed, let’s draw on what we know about the insurance claims process and imagine how such claims might proceed. As possessors of un-artificial intelligence, we’ll spell out at the start that the scenarios we spin are hypothetical.

Our “database” of thirty years of insurance coverage litigation experience offers an organizing principle: Insurance companies often cannot resist the urge to take unfair advantage when an insurance claim happens. 

Imagine: “Insurance Company A rejects coverage on grounds that the claim is an E&O claim, and Insurance Company B rejects coverage on grounds that it’s a cyber claim.”

Imagine an attorney who files a claim seeking defense and indemnity for liability stemming from his citation of legal opinions invented by a chatbot. He provides notice to Insurance Company A, which sold his firm a cyber insurance policy, and Insurance Company B, which provided the firm’s Errors & Omissions coverage. Insurance Company A rejects coverage on grounds that the claim is an E&O claim, and Insurance Company B rejects coverage on grounds that it’s a cyber claim. Insurance companies often point at each other in complex claims. The jaded among us may expect such behavior, but in many cases it is bad faith.

In our A.I. case, the E&O insurance company might point to cyber exclusions in the policy, even though a human error multiplied by an electronic tool is not what most would think of as a cyberattack. The cyber insurance company might argue, conversely, that the attorney-policyholder was not hacked or otherwise cyber-assaulted, or even the victim of a software error (as the chatbot’s construction of fanciful “information” may be in accordance with its current design).

If the basic fact of coverage is indisputable, the claims-avoiding argument often turns to which limit of liability or which deductible applies.

In my experience, insurance companies often choose the policy, exclusion, limit, or retention that suits their financial interests, even if the policyholder intended to buy broad seamless coverage. Such claims behavior over decades gave rise to generally policyholder-friendly rules of insurance policy construction and insurance bad faith law, which courts use to sort through claims disputes.

Insurance Policies Are a Unique Product

Insurance policies uniquely first require the policyholder to perform by paying insurance premiums. The insurance company delays payment of the claim – often excessively — at its option. Policyholders rely on the good faith and fiduciary nature of the insurance company. This disjointed performance of the parties to the insurance policy can lead, particularly in novel matters such as those involving A.I.-multiplied harm, to wrongful claims practices. Policyholders need not permit insurance companies to protect only their financial self-interest. Instead, policyholders can and should insist that policyholder interests in reducing liability and minimizing losses occupy the central position when facing an insurance claim.

Don’t Let A.I. Distract You from Getting Your Claim Paid

The future will permit artificial intelligence tools to help humans make bigger and more severe mistakes. Don’t let your insurance company compound the insult by failing to make a fair payment for a claim.


WILLIAM G. PASSANNANTE is co-chair of Anderson Kill’s Insurance Recovery Group and is a nationally recognized authority on policyholder insurance recovery in D&O, E&O, asbestos, environmental, property, food-borne illness, and other insurance disputes. Clients include global corporations and educational and governmental institutions.