AI liability, insurance and the challenge for boards of directors

As AI adoption expands, boards and C-suites raise questions of accountability, liability, insurability and broader risk management. In 2026, concern shifts from theory to practice as legal, regulatory and technical frameworks evolve.

Where AI responsibility remains unclear

Recent cases have challenged boards and senior leadership teams by exposing unclear responsibility for AI-related risk, including:

• Employment discrimination: Job seekers allege AI screening tools discriminate based on race, age and disability, raising questions about whether liability rests with the employer that used the tool or the vendor that designed the algorithm
• AI-generated misinformation and defamation: Lawsuits allege AI hallucinations wrongly identified people as convicted criminals, college dropouts or having been involved in scandals, highlighting the unclear liability between AI creators, trainers and users
• Copyright infringement: Content creators allege AI companies unlawfully used copyrighted material for training, while AI firms cite fair use, leaving liability unclear among the model developer, the developers of the scraped data and end-users
• AI-driven mental health harm: Allegations that chatbots contribute to death by suicide raise novel questions about whether AI is a product with liability resting on the developer or responsibility resting on users
• AI-powered autonomous discounts and rebates: Courts rule that companies cannot ignore discounts or rebates that are unintentionally generated by their AI tools
• AI-powered autonomous driving: Fatal incidents involving AI-driven vehicles raise unresolved questions of responsibility

AI transforms risk

WTW’s Sonal Madhok and Dr. Anat Lior, assistant professor at Drexel University’s Thomas R. Kline School of Law, and Yale University’s Jackson School for Global Affairs, recently wrote that AI is transforming risk profiles across industries. AI-related risks largely are covered implicitly under traditional insurance policies through “silent AI” coverage, where AI may not be specifically referenced. While current policies may cover AI incidents such as cyber or data liability even if they don’t specifically mention AI, such coverage creates ambiguity. Insurers increasingly are clarifying coverage for AI through endorsements that affirm coverage for certain AI risks or adding exclusions to standard policies to avoid unanticipated exposure.

Fragmented coverage

Companies often rely on a patchwork of insurance policies to cover AI risks, without a single policy covering all potential exposures. AI losses may fall across cyber, general liability and other lines, creating gaps in coverage. Insurers expect the AI insurance market to grow to approximately $4.7 billion in premiums by 2032. Carriers have introduced specialized AI endorsements, with some cyber policies explicitly covering or excluding AI-driven events such as data poisoning or AI-generated content and some standalone offerings to address specific needs.

Dr. Lior’s publication, Insuring the AI Age, in the Connecticut Insurance Law Journal, explains how organizations currently address AI risks through existing insurance policies, such as the following 12 common policy types:

1. Cyberinsurance: Covers data breaches, AI-aided hacks and privacy violations. Requires a breach trigger. Typically excludes unauthorized AI use, AI-generated content and the insured’s own data loss. Example: Hackers gain access to proprietary customer data through AI
2. Tech errors and omissions: Covers errors in technological services/products, including AI systems. Requires clear negligence. Typically excludes high-risk AI apps, lack of oversight and injuries/damages. Example: AI trading algorithm causes client losses
3. Employment practices liability insurance: Covers discrimination from AI in HR or customer service settings. Limited to covered groups. Typically excludes bias sub-limits. Example: AI hiring tool discriminates against older applicants
4. Professional indemnity: Covers errors by professionals using AI tools. Assumes human oversight. Coverage under autonomous AI remains unclear. Typically excludes unsupervised or unapproved AI use. Example: Physician misdiagnoses due to AI tool
5. General liability: Covers injury and damage from AI systems, including defamation. Typically excludes intangible harm, data-only events, and financial or cyber loss. Example: AI robot injures visitor or customer
6. Workers compensation: Covers employee injuries from AI-operated equipment. Only covers employees. Provides limited benefits. Few exclusions, but premiums may be subject to adjustment. Example: Factory worker injured by AI-controlled robot
7. Intellectual property (IP) liability: Covers infringement from AI-generated content. Typically excludes training-data IP claims, generative AI and intentional acts. Example: AI-generated image infringes copyright
8. Property insurance: Covers physical damage from AI failures. Typically excludes downtime, cyber, data and software loss. Example: AI system causes explosion and property damage
9. Crime insurance: Covers theft and fraud, including AI-enabled scams. Sub-limits for social engineering and non-fraud losses. Typically excludes certain deep-fake fraud and insider fraud. Example: Deep-fake voice scam leads to wire transfer
10. Directors’ and officers’ liability: Covers executive liability for AI oversight failures. Specific exclusions are still emerging with scrutiny of AI governance. Example: Shareholders sue board over failed AI project
11. Product liability: Covers injury and damage from defective AI-enabled products. Definition of AI as “product” remains unclear. Typically excludes financial loss and noncompliance with safety standards. Example: A self-driving car’s AI fails and causes a crash
12. Media liability: Covers defamation, privacy and IP from AI content. Often requires editorial review. Typically excludes intentional acts, unreviewed AI content and patent issues. Example: AI-generated article fabricates content

Actions to address AI risk

Effective leaders currently take six actions to address AI risk:

• Stay up to date on the evolving AI risk landscape and ensure insurance policies cover the necessary perils without surprising exclusions. Be aware of potential coverage gaps or ambiguities, particularly for novel AI applications outside established domains
• Track regulatory and legislative developments as well as insurer responses. Regulatory shifts can rapidly alter liability exposures and policy requirements
• Implement strong AI governance, following processes such as the updated AI governance model, which includes AI risk, from the National Association of Corporate Directors (NACD) as reported by Dylan Sandlin, Helmuth Ludwig and Dr. Benjamin van Giffen
• Reassess traditional risk models (which may not fully capture the risks associated with AI as new technologies and use cases emerge) and evaluate whether current coverage is sufficient or bespoke solutions are warranted
• Monitor litigation and claim trends to ensure adequate coverage as court decisions increasingly influence underwriting, policy design and claim outcomes
• Prepare for newly converging technologies such as quantum computing and robotics, which will further complicate systemic risk, supply chains and regulatory response

As AI reshapes risk and liability, effective leaders move beyond fragmented coverage and unclear accountability. They understand proactive governance, informed insurance decisions and continuous monitoring are critical to managing AI risk in a changing legal and regulatory landscape.

A version of this article originally appeared on Forbes on April 13, 2026.