AI Needs Its Own Risk Class: Lockton Re

Why It Matters

AI is reshaping liability across every sector, and existing commercial policies leave critical gaps; defining a separate AI risk class will align coverage with emerging exposures and protect both insurers and policyholders.

Key Takeaways

• AI amplifies cyber attacks via deep‑fakes and phishing.
• Traditional E&O policies miss probabilistic AI failure scenarios.
• CGL insurers lack AI underwriting, creating coverage gaps.
• D&O exposure rises from AI governance failures and “AI washing.”
• Affirmative AI policies provide all‑risk, model‑specific coverage.

Pulse Analysis

The rapid diffusion of generative models and large‑language‑AI tools has turned artificial intelligence into a cross‑industry catalyst for new liability exposures. Insurers, historically organized around discrete lines such as cyber or professional liability, now confront claims that blend technical malfunction, data bias, and reputational harm. By carving out AI as its own risk class, carriers can develop standardized definitions, pricing metrics, and policy language that reflect the probabilistic nature of model behavior, rather than retrofitting legacy clauses that often miss the nuance of AI‑driven loss.

Current commercial lines reveal significant coverage gaps. Cyber policies may only cover AI‑related incidents when a traditional breach triggers the loss, while E&O contracts remain anchored to deterministic software failures, leaving algorithmic errors and hallucinations uninsured. Casualty and D&O policies lack clear AI exclusions or endorsements, exposing firms to governance claims and "AI washing" accusations. The report underscores the need for systemic underwriting approaches that assess portfolio‑level concentration on shared foundation models, recognizing that a single compromised model can generate simultaneous claims across disparate sectors.

Affirmative AI coverage is emerging as a pragmatic response, offering all‑risk, model‑specific protection that evaluates each deployment on factors such as industry context, version control, and data provenance. This bespoke underwriting enables insurers to price risk more accurately and provides policyholders with clearer assurances against AI failures that fall outside traditional perils. As regulatory scrutiny intensifies and AI becomes integral to core business processes, the establishment of a dedicated AI risk class will likely become a cornerstone of modern commercial insurance portfolios.