AI is transforming the world, and it's just the beginning. While we may marvel at the increasing capabilities of frontier models, they are creating risks and liabilities that we don't fully understand yet from an insurance perspective. Generative AI-related lawsuits in the US grew 978% from 2021 to 2025. Let that soak in.1
Generally, insurance policies have covered “AI liability” risks mostly through existing coverages, such as cyber, technology errors and omissions, product liability, and commercial general liability (termed as the “silent AI” coverage). However, with the growing applications of AI, the insurance industry is realizing that there are substantial gaps with this approach. Increasingly, the industry is starting to narrow coverage for AI liability. The Insurance Services Office (the body that standardizes US policy language) issued a generative AI exclusion for commercial general liability policies.2
Carriers, too, are increasingly calling out AI liability as an exclusion. Berkley, for instance, has rolled out an “Absolute AI Exclusion.” There is a real danger that as companies renew current policies that new gaps exist that need to be addressed. An example would be concerning models, agents and data pipelines that are compromised. In 2025, researchers uncovered malicious machine learning models (nullifAI) on a public platform3 and found a popular AI tool that could be tricked into following hidden instructions slipped into its input — effectively letting an outside attacker seize control of the system it was running on.4 Such incidents would be likely excluded from current policies not specifically addressing AI liabilities. One area that I've been really interested in is the growing number of autonomous AI systems and all the associated risks that will certainly emerge.
AI-specific products are slowly emerging to fill that gap, such as Munich Re's aiSure, Armilla, and Testudo, among others. Still, the fast pace of change is leaving plenty of gaps in liability coverage for companies adopting AI into their workflow or product.
As someone who ran an AI company and launched an AI agent-based product, Addy — a document workflow automation agent for healthcare clinics — I understand the growing risks AI is creating firsthand. Now, wearing the insurance provider hat, I'm in a unique position to help other start-up and mid-cap executives navigate this shifting landscape and mitigate AI liability in their own companies.