AI risks make some insurers wary of corporate liability

The insurance industry is growing cautious about the risks associated with artificial intelligence (AI), according to a report by the RAND Corporation. While AI-related harms have begun to emerge, they are not easily categorized within existing insurance policies. These risks include incorrect or misleading outputs, deepfakes, privacy breaches, intellectual property disputes, fraud, product defects, and discriminatory decisions.

The rapid adoption of AI by businesses has created a fragmented market for insuring AI usage, particularly in the United States.

Companies are seeking insurance to protect themselves from financial losses due to unreliable, deceptive, and immoral AI agents, but insurance companies are hesitant to accept this risk. If AI becomes uninsurable, the AI industry may need to adjust its ambitions and sales targets, while corporate customers may delay AI projects to ensure they comply with their fiduciary responsibilities.

The report notes that some insurance carriers, such as W. R. Berkley, have already introduced exclusions in their directors and officers, errors and omissions, and fiduciary liability insurance products to exclude coverage for any actual or alleged use, deployment, or development of AI. In January 2026, Verisk/ISO introduced optional language for carriers to exclude bodily injury, property damage, and other harms arising from generative AI in their standardized forms.

Despite these exclusions, new and existing insurers are filling coverage gaps by developing their own risk calculations. The Artificial Intelligence Incident Database (AIIDB) lists 713 incidents of harm or near-harm caused by generative AI, covering various categories such as misinformation, deepfakes, hallucination, harmful content, agentic and autonomous failure, privacy leaks, bias, copyright and IP issues, and wrongful AI attribution.

Additionally, there are approximately 250 U.S. lawsuits related to AI, primarily concerning copyright and IP, as well as privacy, fraud, negligence, product liability, discrimination, and contracts/trade secrets. Various laws in U.S. states cover AI-related intimate images, AI-generated child sexual abuse material, automated decision-making, AI in political ads, and other related rules that could affect businesses deploying AI.

While the AI industry may temporarily face a pause in development due to insurers' uncertainty about pricing the risks, RAND Corporation advises that businesses continue to deploy AI. To address the market gaps, RAND suggests that policy researchers, brokers, carriers, and reinsurers should develop a common taxonomy to track AI incidents and claims.

State regulators should also introduce an AI Coverage Notice to clarify which risks are covered and which are not. The report believes that understanding and pricing these risks will eventually allow AI insurance coverage to move beyond being a specialty product.