• Atlanta
• Boston
• Dallas
• Ft. Lauderdale
• London
• Minneapolis
• New York
• Oakland
• Philadelphia
• Washington, DC

by Marsheldondria C. Johnson

AI is being used to manipulate photos of damage and documents, and even to generate images of items and losses that never existed. Insurers are countering with AI-detection technology, but are policies keeping up with the changing times?

According to the 2026 Verisk State of Insurance Fraud study, 36% of consumers say they would consider digitally altering a claim image or document even if it broke insurer rules. 98% of insurance carriers agree that AI editing tools are fueling an increase in digital insurance fraud. Is your SIU team equipped for the challenge?

It starts with the contract. Most carriers still rely on standard concealment/fraud and misrepresentation clauses to address AI-fabricated claim evidence. Yes, those clauses do the job, since a fake photo is simply another form of material misrepresentation. But as this form of technology rapidly transforms our lives and the insurance industry, we must keep in mind that generic fraud language did not contemplate generative AI. As such, these provisions could fall short in the future when arguments over what misrepresentation covers arise when fraud consists of algorithmically generated pixels rather than a material misrepresentation typed into a form.

Something worth pondering: the addition of explicit disqualifying language naming AI-generated or AI-altered images and documentation in the claims-reporting and proof-of-loss provisions. This accomplishes two things:

1. Deterrence- A named prohibition puts claimants on notice before they submit anything, rather than after.
2. Litigation posture- If a coverage dispute follows, explicit language gives counsel a cleaner, less-arguable basis to deny or rescind than a general fraud clause stretched to fit a new fact pattern.

To illustrate how deleterious these schemes can be to the insurance industry, see the results of our internal AI experiments below.

Note these images were created within a matter of seconds.

Example of AI-generated estimate manipulation:

Original Estimate:

AI-Manipulated Estimate

Example of an AI-generated damage exaggeration:

Original Photo		AI-Generated Photo

Top 3 Tools for Your SIU Team:

1. Metadata analysis- examining file creation data, editing history, and source signatures that AI-generated images often strip or fake.
2. AI-recognition software-detection tools trained to catch the artifacts (i.e., mismatched shadows, repeated pixel patterns, lighting inconsistencies) that give away synthetic images.
3. In-person inspection-when metadata or software flags a submission as suspect, an on-site adjuster visit remains the most reliable check available.

The Lowdown:

The rise of AI technology is quickly integrating into every facet of our lives, including our insurance claims. To ensure we are equipped to tackle the issues it can present, we must re-evaluate our lines of defense. Existing fraud clauses technically reach AI-fabricated claim evidence, but they weren't written with AI in mind, so the practical move for insurance carriers is to contemplate tightening policy language to explicitly name AI-generated/altered submissions as disqualifying, while providing your SIU team with additional support with stronger proof-of-loss requirements and detection tools such as metadata analysis, AI-recognition software, and in-person inspections to keep pace.

The opinions expressed are those of the authors and do not necessarily reflect the views of the firm or its clients. This article is for general information purposes and is not intended to be and should not be taken as legal advice.