Artificial intelligence (AI) has rapidly transitioned from an experimental technology to a core component of corporate strategy, professional service delivery, and operational decision-making. Company boards are overseeing AI adoption, executives are disclosing AI strategy to investors, professionals are relying on AI-assisted outputs, and cyber criminals are using AI to scale attacks. As a result, AI has become a meaningful driver of litigation, regulatory scrutiny, and insurance claims.
From an insurance standpoint, AI does not represent a new standalone peril. Instead, it reframes traditional liability exposures under Directors & Officers (D&O), Errors & Omissions (E&O), and Cyber policies. Coverage frequently exists today, but it is increasingly tested by exclusions, definitions, sublimits, and underwriting assumptions. This makes it critical that retail agents understand how AI-related claims trigger coverage, where coverage boundaries lie, and how insureds can preserve protection as AI risk changes.
D&O COVERAGE THEORY: AI AS A “WRONGFUL ACT”
Most D&O policies define a “Wrongful Act” broadly to include actual or alleged breaches of duty, misstatements, omissions, or neglect. Decisions involving AI, such as whether to adopt AI tools, how to govern them, how to oversee risk, and how to disclose AI capabilities, are quintessential board- and management-level actions. Accordingly, AI-related claims typically fit squarely within the D&O insuring agreement and are often characterized as silent D&O exposure.
PRIMARY AI-DRIVEN D&O CLAIMS: SECURITIES AND DISCLOSURE LITIGATION
The most significant D&O exposure to date involves allegations that companies overstated or mischaracterized AI capabilities, commercialization readiness, or financial impact. These claims frequently arise from earnings calls, SEC filings, IPO or SPAC disclosures, and public marketing statements, often resulting in stock-drop litigation following performance shortfalls.
DERIVATIVE ACTIONS: FAILURE OF OVERSIGHT
Shareholders increasingly allege that boards failed to implement appropriate AI governance frameworks, failed to monitor known risks such as bias or hallucinations, or failed to ensure regulatory compliance. These claims commonly implicate Side A coverage where indemnification is unavailable, increasing the importance of limits and Side A structure.
REGULATORY INVESTIGATIONS
Regulators have signaled heightened scrutiny of AI disclosures and governance. D&O policies often respond to defense costs associated with Securities and Exchange Commission (SEC), Department of Justice (DOJ), Federal Trade Commission (FTC), and state-level investigations, subject to the policy’s definition of “Claim” and investigation-related wording.
COVERAGE BOUNDARIES AND EMERGING RISKS
D&O policies generally do not cover product liability, bodily injury or property damage, professional service errors, or cyber events. AI-related matters frequently give rise to allocation disputes among D&O, E&O, and Cyber policies. While explicit AI exclusions are not yet standard in D&O, some insurers are proposing broad exclusionary language, making policy review and negotiation increasingly important.
E&O COVERAGE THEORY: AI AS A PROFESSIONAL TOOL
Courts and insurers generally treat AI as a tool used in professional services rather than an autonomous decision-maker. Responsibility for AI-assisted outputs remains with the professional or firm using the technology. As a result, E&O is typically the primary line of coverage for AI-driven professional mistakes resulting in financial loss.
COMMON AI-DRIVEN E&O CLAIM SCENARIOS
E&O claims commonly allege incorrect or flawed AI outputs, hallucinated or inaccurate advice, biased or discriminatory results, failure to supervise AI tools, and negligent reliance on automated systems without sufficient human review.
WHERE E&O COVERAGE CAN ERODE
Coverage often turns on whether AI-assisted activities fall within the policy’s definition of professional services. Risk increases when internal tools are repurposed for client use, when AI platforms are bundled with services, or when outputs are delivered without human validation. Broad technology, cyber, data, or AI-specific exclusions can materially erode coverage if not carefully addressed.
Examples of prohibitive exclusions include:
- Absolute AI Exclusions: Broad exclusions remove coverage for claims arising out of any use or deployment of artificial intelligence, often sweeping in human negligence.
- Generative AI Exclusions: These exclude claims involving AI-generated text, imagery, audio, or data, eroding coverage for routine professional tasks.
- Technology and Algorithm Exclusions: Older tech exclusions are now interpreted to bar claims involving AI models or automated systems.
- Autonomous Decision-Making Exclusions: Claims tied to automated or insufficiently supervised AI decisions may be excluded.
- Governance and Supervision Exclusions: Some endorsements remove coverage for failure to implement AI policies or oversight.
CYBER COVERAGE THEORY: AI AS A COVERED ATTACK VECTOR
Cyber insurance is currently the most resilient line with respect to AI risk. Policies are generally technology-neutral, meaning that if a loss would be covered when caused by a human, it is typically covered when AI is used to execute or enhance the attack. Insurers view AI as a threat multiplier rather than a new excluded peril.
AI-DRIVEN CYBER LOSS SCENARIOS
Common AI-driven cyber claims include deepfake-enabled social engineering, AI-enhanced phishing and credential theft, data leakage through AI tools without traditional hacking, and attacks targeting AI models or training data. Coverage often depends on definitions, sublimits, and compliance with control requirements.
PRACTICAL CYBER COVERAGE WATCHPOINTS
Key cyber coverage risks involve narrow social engineering definitions, hidden sublimits, and business interruption triggers requiring system outages rather than data or model corruption. As AI increases attack sophistication, cyber underwriting is increasingly focused on behavioral and process controls in addition to technical safeguards.
BOTTOM LINE
AI isn’t creating entirely new liabilities; it’s accelerating, amplifying, and complicating the ones your clients already have. The real risk isn’t whether coverage exists, but whether it responds the way you expect when an AI-driven claim hits. Subtle shifts in policy language, exclusions, definitions, and underwriting assumptions are already determining which claims get paid, and which don’t.
For retail agents, this is where experience and access matter. AI-related risks rarely fit neatly into a single policy. They require coordinated structuring across D&O, E&O, and Cyber, along with careful negotiation to avoid silent gaps and unintended exclusions. These are not “standard” placements, and carriers are evolving their approach in real time.
CRC Specialty’s professional lines specialists work directly with leading markets to provide coverage that aligns with how AI exposures actually develop, helping you anticipate gaps, position submissions effectively, and secure solutions that hold up when tested.
As AI continues to reshape liability, the agents who win will be the ones who understand how to translate innovation into insurable risk - and those who have the right wholesale partner to navigate it. Reach out to your CRC Specialty broker to make sure your clients’ coverage keeps pace with how their risk is changing.
CONTRIBUTOR
- Jenny Finet is a Professional Lines Broker with more than 20 years of experience. She is based in CRC’s Chicago office and holds both CPCU and RPLU designations.
END NOTES
- AI Cyber Attack Statistics 2025, Trends, Costs, Defense, DeepStrike, October 10, 2025. https://deepstrike.io/blog/ai-cyber-attack-statistics-2025
- 40+ Alarming Enterprise AI Breach & Security Risk Statistics, Index, February 5, 2026. https://www.index.dev/blog/enterprise-ai-security-risk-statistics