Insurance
Home›Insurance›Industry & Deals›Insurers diverge on AI coverage as deepfakes and misin…
Insurers diverge on AI coverage as deepfakes and misinformation drive claims
In a review of 713 generative AI incidents since ChatGPT’s 2022 launch, RAND found 84% involved misinformation or deepfakes.
Insurers remain split on how to handle coverage for harms linked to artificial intelligence, as a RAND Corporation research report highlights inconsistent approaches ranging from exclusions to affirmative coverage, or silence, depending on the type of loss.
According to the report, which draws on an AI Incident Database and state laws plus rate and form filings, 713 generative AI incidents were tracked since ChatGPT’s November 2022 launch. RAND found 599 incidents, or 84%, involved misinformation or deepfakes, while other harms included hallucinations and factual errors, harmful content, and agentic or autonomous failures.
The report also distinguishes between incident patterns and litigation. Of 249 U.S. generative AI lawsuits reviewed, RAND found 150, or 60%, focused on intellectual property or training-data disputes against AI developers, rather than direct harms from model outputs, with privacy and surveillance, fraud and deception, and product liability claims taking smaller but growing shares.
RAND found most enacted state AI laws it examined targeted harmful synthetic content, noting that 33 states have passed 62 laws addressing nonconsensual intimate images and child sexual abuse material. The report says insurers do not treat AI as a single peril, because the same model can produce different categories of loss based on how it is trained, deployed, and used, and some major carriers have moved to exclude AI losses broadly, including through standardized language developed by Verisk/ISO.