AI Corporate Liability – Insurance Insurers Wary of AI Risk Coverage

The insurance industry faces an emerging crisis: artificial intelligence adoption is creating liability gaps that traditional policies don't adequately address, and carriers are increasingly balking at coverage. A new RAND Corporation analysis reveals that AI-related harms are proliferating—including incorrect or misleading outputs, deepfakes, privacy violations, intellectual property disputes, fraud, product defects, and discriminatory decisions—yet the insurance market lacks the actuarial data to properly price these novel risks. The problem has reached critical mass. In January 2026, Verisk/ISO—whose standardized forms appear in over 80 percent of US property and casualty policies—introduced optional AI exclusions. This move signals that major underwriters are retreating from AI-related coverage rather than attempting to quantify and insure it. Corporate liability policies that previously seemed comprehensive are now riddled with gaps. The core issue isn't hard to understand: AI failures operate at scales and speeds that traditional insurance models can't accommodate. A single hallucination in a language model deployed company-wide could create liabilities across multiple policy categories simultaneously. Discriminatory AI systems might trigger both employment liability and product liability claims. Privacy violations through data misuse affect countless individuals at once. Insurance actuaries, accustomed to historical loss data and predictable claim patterns, are facing an entirely new risk landscape with minimal historical precedent. Companies deploying AI systems are increasingly finding themselves without adequate insurance protection. Directors and officers operating under the false assumption that traditional D&O policies cover AI-related liability face unrecognized exposure. The RAND report essentially argues that the insurance industry needs better data collection, more sophisticated modeling, and perhaps entirely new policy categories designed specifically for AI deployment. Until the market develops adequate coverage mechanisms, corporations deploying AI systems are taking on unmeasured and uninsurable risks. The gap between deployment velocity and insurance market maturity is widening—a potentially dangerous mismatch for enterprises betting their operations on machine learning systems.