Insurers Tighten AI Coverage as Litigation Expands
Insurance carriers are restricting coverage for artificial intelligence risks through broad exclusions and endorsements, even as courts have yet to clarify how these provisions will hold up legally. The moves reflect a surge in AI-related litigation spanning copyright disputes, product liability claims, privacy violations, and algorithmic bias allegations.
Some insurers have adopted what they call "absolute" AI exclusions in directors and officers, errors and omissions, and fiduciary liability policies. These provisions can block claims arising from AI use, related disclosures, regulatory violations, or alleged breaches of AI-specific laws. Other carriers offer narrower exclusions targeting generative AI systems-those producing text, images, audio, or synthetic data-or they provide affirmative AI-specific coverage options.
The Litigation Driving Coverage Changes
Recent cases illustrate the breadth of AI-related claims. Bartz v. Anthropic, a copyright dispute tied to training large language models, reportedly settled for $1.5 billion. Raine v. OpenAI, Inc. involves allegations of real-world harm from AI deployment. Reddit, Inc. v. Anthropic PBC challenges the use of scraped data for training, while Chegg, Inc. v. Google LLC raises antitrust questions about proprietary data in AI systems.
Claims also include discrimination and algorithmic bias, as in Mobley v. Workday, Inc., and securities class actions tied to AI capability disclosures, such as D'Agostino v. Innodata Inc. This range of legal theories has prompted insurers to broaden exclusionary language across policy types.
How Exclusion Language Varies
Definitions matter. One insurer defines artificial intelligence broadly as any machine-based system generating predictions, content, or decisions from input data-a definition that can capture tools used in business for years. Other carriers focus narrowly on generative AI systems, sometimes naming specific tools like ChatGPT, Bard, Midjourney, or DALL-E.
The Insurance Services Office has introduced optional endorsements for commercial general liability policies. These exclude coverage for bodily injury, property damage, personal and advertising injury, and products liability claims arising from generative AI.
Courts May Limit Exclusion Scope
Legal standards for policy interpretation could constrain how broadly these exclusions apply. Courts typically construe coverage provisions broadly and exclusions narrowly. An insurer may still have a duty to defend claims that mix AI-related and non-AI allegations if any portion is potentially covered.
Coverage may also apply when an alleged injury could occur independently of AI use or when claims do not clearly fall within the exclusion language. Courts have rejected exclusions that eliminate the fundamental protections a policy is meant to provide, treating such provisions as illusory coverage.
Earlier policies issued before AI-specific exclusions were added may still offer protection. Changes in policy language over time influence how courts interpret older forms, potentially preserving coverage under pre-exclusion policies.
For AI for Insurance professionals and AI for Legal teams, understanding these policy dynamics is essential as litigation continues to develop and courts begin interpreting exclusion language.
Your membership also unlocks:
