Insurance Policy Updates for AI-Related Risks
As generative artificial intelligence (AI) becomes increasingly integrated into business operations, insurers are revising policy language to address the novel risks it presents. This shift occurs alongside a rise in litigation related to the development, deployment, and transparency of AI, prompting a greater use of exclusions and endorsements within liability policies.
Insurers have begun implementing provisions to either restrict or remove coverage for AI-related risks, but the effectiveness and legal standing of these exclusions are still evolving. Some insurance providers have opted for tighter coverage through specific endorsements, while others have proposed policies that actively cover AI-related issues. Courts are currently determining the impact of these exclusions, especially in cases where coverage may be significantly restricted.
Recently, there has been a surge in litigation involving a diverse array of AI-related claims. These include disputes over copyright and intellectual property, such as the Bartz v. Anthropic case, which reportedly settled for $1.5 billion. Cases addressing product liability and negligence, like Raine v. OpenAI, Inc., have also emerged, focusing on allegations of tangible harm.
Privacy and data usage claims are evolving, exemplified by Reddit, Inc. v. Anthropic PBC, which questions the use of collected data for AI training. Antitrust claims, including Chegg, Inc. v. Google LLC, target the use of proprietary data in AI frameworks. Other lawsuits involve discrimination and algorithmic bias, such as Mobley v. Workday, Inc., and securities class action suits concerning AI capability disclosures, like D’Agostino v. Innodata Inc.
In response to extensive litigation, insurers are adopting broader exclusionary language. Some insurance firms have introduced “absolute” AI exclusions in directors and officers, errors and omissions, and fiduciary liability policies. These clauses may address claims linked to AI usage, relevant disclosures, and compliance with regulatory mandates or legal breaches.
The Insurance Services Office (ISO) has rolled out optional endorsements for commercial general liability policies. These entail exclusions for bodily injury, property damage, personal and advertising injury, and products and completed operations liability when claims stem from generative AI.
Legal standards pertinent to policy interpretation continue to play a crucial role. Generally, courts interpret coverage provisions broadly and exclusions narrowly. Insurers might still have an obligation to defend claims involving both AI-related and non-AI-related allegations. Additionally, policies issued before the advent of AI-specific exclusions might still offer coverage, as changes in policy language can influence court interpretations of earlier documents.